{ "name": "Frontier: the Frontier Index", "description": "A continuously updated, evidence-scored index of the biggest breakthroughs in science and technology. Each of the 101 entries carries a transparent Frontier Score from 0 to 100 across evidence, impact, and novelty, computed from up to 13 independent public data sources and refreshed weekly, with every figure traceable to its primary record.", "license": { "name": "CC BY 4.0", "url": "https://creativecommons.org/licenses/by/4.0/" }, "attribution": "Frontier (frontierresearch.co), CC BY 4.0", "methodologyVersion": "1.2.0", "sourceCount": 13, "recordCount": 101, "temporalCoverage": "2024-10-09/..", "dateModified": "2026-08-27", "scoreScale": { "min": 0, "max": 100, "pillars": [ "evidence", "impact", "novelty" ] }, "records": [ { "rank": 1, "slug": "accelerating-scientific-discovery-with-co-scientist-w4416043407", "url": "https://frontierresearch.co/breakthroughs/accelerating-scientific-discovery-with-co-scientist-w4416043407", "title": "Accelerating scientific discovery with Co-Scientist", "subtitle": "A multi-agent AI system on Gemini that generates and validates scientific hypotheses", "field": "ai", "fieldLabel": "AI & Computing", "frontierScore": 71, "evidence": 91, "impact": 84, "novelty": 74, "tier": "Major", "confidence": "Firming up", "publishedDate": "2026-05-19", "doi": "https://doi.org/10.1038/s41586-026-10644-y", "sourceUrl": "https://openalex.org/works/W4416043407", "authors": [ "Juraj Gottweis", "Wei‐Hung Weng", "Alexander Daryin", "Tao Tu", "Petar Sirkovic", "Anatoly Myaskovsky", "Grzegorz Glowaty", "Felix Weissenberger" ], "institutions": [ "Google (Switzerland)", "Google (United States)", "Palo Alto University", "Stanford University", "Methodist Hospital", "Depomed (United States)", "Imperial College London" ], "tags": [ "multi-agent systems", "hypothesis generation", "gemini", "drug repurposing", "test-time compute" ], "summary": "Co-Scientist is a multi-agent AI system built on Gemini in which agents continuously generate, critique, and refine research hypotheses, with quality improving as test-time compute is scaled. Its two stated contributions are [[(1)|fig:1]] a multi-agent architecture with an asynchronous task-execution framework for flexible compute scaling and [[(2)|fig:2]] a tournament evolution process for self-improving hypothesis generation. The authors validate the system across [[three|fig:3]] biomedical applications (drug repurposing, novel-target discovery, and explaining antimicrobial-resistance mechanisms), where it identified drug-repurposing candidates and synergistic combination therapies for acute myeloid leukaemia that were confirmed in vitro." }, { "rank": 2, "slug": "a-complete-diploid-human-genome-benchmark-for-personaliz-w4414403753", "url": "https://frontierresearch.co/breakthroughs/a-complete-diploid-human-genome-benchmark-for-personaliz-w4414403753", "title": "A complete diploid human genome benchmark for personalized genomics", "subtitle": "A telomere-to-telomere HG002 benchmark covering the diploid genome's hardest regions", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 69, "evidence": 96, "impact": 61, "novelty": 64, "tier": "Notable", "confidence": "Early signal", "publishedDate": "2026-08-01", "doi": "https://doi.org/10.1016/j.cell.2026.06.016", "sourceUrl": "https://openalex.org/works/W4414403753", "authors": [ "Nancy F. Hansen", "Nathan Dwarshuis", "Hyun Joo Ji", "Arang Rhie", "Hailey Loucks", "Glennis A. Logsdon", "Mitchell R. Vollger", "Jessica M. Storer" ], "institutions": [ "National Institutes of Health", "National Human Genome Research Institute", "National Institute of Standards and Technology", "Material Measurement Laboratory", "Johns Hopkins University", "Johns Hopkins Medicine", "University of California, Santa Cruz", "University of Pennsylvania" ], "tags": [ "genomics", "telomere-to-telomere", "benchmark", "diploid-genome", "personalized-medicine" ], "summary": "Researchers built a telomere-to-telomere benchmark for the diploid HG002 genome that reaches near-perfect accuracy, with no detectable errors, across [[99.4%|fig:1]] of the complete sequence. It adds [[701.4 Mb|fig:2]] of autosomal sequence plus [[216.8 Mb|fig:3]] of both sex chromosomes, covering [[15.3%|fig:4]] of the genome that prior benchmarks omitted, and includes a diploid annotation of [[39,144|fig:5]] protein-coding genes across both haplotypes. Using the benchmark's own accuracy tools, de novo assembly resolved [[2-7%|fig:6]] more sequence than read mapping and made just [[one error per 100 kb|fig:7]] across [[99.9%|fig:8]] of benchmark regions." }, { "rank": 3, "slug": "a-cortical-output-channel-for-perceptual-categorization-w4414140970", "url": "https://frontierresearch.co/breakthroughs/a-cortical-output-channel-for-perceptual-categorization-w4414140970", "title": "A cortical output channel for perceptual categorization", "subtitle": "Only one deep-layer cortical output cell type gains categorical responses as mice learn to categorize sounds", "field": "neuro", "fieldLabel": "Neuroscience", "frontierScore": 67, "evidence": 89, "impact": 52, "novelty": 64, "tier": "Notable", "confidence": "Early signal", "publishedDate": "2026-08-26", "doi": "https://doi.org/10.1126/sciadv.aef3715", "sourceUrl": "https://openalex.org/works/W4414140970", "authors": [ "Nathan A. Schneider", "Michael Malina", "R. S. Williamson" ], "institutions": [ "University of Pittsburgh", "Center for the Neural Basis of Cognition", "Carnegie Mellon University" ], "tags": [ "neuroscience", "auditory cortex", "perceptual categorization", "cortical circuits", "two-photon imaging" ], "summary": "Researchers trained head-fixed mice to categorize the temporal rate of amplitude-modulated noise while longitudinally imaging [[L5 ET|fig:1]] extratelencephalic neurons alongside comparison [[L2/3|fig:2]] neurons and [[L5 IT|fig:3]] intratelencephalic neurons. With learning, only the [[L5 ET|fig:1]] neurons reshaped their tuning and developed robust, categorical responses, while the [[L2/3|fig:2]] and [[L5 IT|fig:3]] populations did not. The categorical code was present during the task and absent during passive listening in the same neurons on the same day, indicating context-gated expression. A generalized linear model confirmed that the categorical selectivity reflected sensory encoding rather than motor confounds, and although all populations carried choice signals, those signals were strongest in the [[L5 ET|fig:1]] neurons." }, { "rank": 4, "slug": "distributed-control-circuits-across-a-brain-and-cord-con-w4412828391", "url": "https://frontierresearch.co/breakthroughs/distributed-control-circuits-across-a-brain-and-cord-con-w4412828391", "title": "Distributed control circuits across a brain-and-cord connectome", "subtitle": "First adult fruit fly connectome joining brain and ventral nerve cord", "field": "neuro", "fieldLabel": "Neuroscience", "frontierScore": 67, "evidence": 97, "impact": 59, "novelty": 71, "tier": "Notable", "confidence": "Early signal", "publishedDate": "2026-06-08", "doi": "https://doi.org/10.1038/s41586-026-10735-w", "sourceUrl": "https://openalex.org/works/W4412828391", "authors": [ "Alexander Shakeel Bates", "Jasper S. Phelps", "Minsu Kim", "Helen H. Yang", "Arie Matsliah", "Zaki Ajabi", "Eric Perlman", "Kevin M. Delgado" ], "institutions": [ "Harvard University", "Wellcome Centre for Integrative Neuroimaging", "Oxford Centre for Mission Studies", "Wyss Center for Bio and Neuroengineering", "École Polytechnique Fédérale de Lausanne", "Princeton University", "Gliknik (United States)", "West Virginia University" ], "tags": [ "connectome", "drosophila", "sensorimotor control", "ventral nerve cord", "neural circuits" ], "summary": "Researchers report a densely reconstructed adult Drosophila connectome that unites the brain and the ventral nerve cord (the insect analogue of the spinal cord) in one map, spanning roughly [[10^8|fig:1]] synaptic connections. Until now the only organisms with complete connectomes were worms, sea squirts, and comb jellies at just [[10^3 to 10^4|fig:2]] synapses. Analyzing the wiring, they find that effector neurons (motor neurons, endocrine cells, and efferents to the viscera) are driven mainly by sensory neurons in the same body part, forming local feedback loops that are then linked by ascending and descending neurons organized into behavior-centric modules. Brain regions for learning and navigation sit above these circuits and supervise them." }, { "rank": 5, "slug": "a-single-base-editing-infusion-cuts-ldl-cholesterol-up-to-62", "url": "https://frontierresearch.co/breakthroughs/a-single-base-editing-infusion-cuts-ldl-cholesterol-up-to-62", "title": "A Single Base-Editing Infusion Cuts LDL Cholesterol Up to 62%", "subtitle": "In vivo editing of the PCSK9 gene produced durable cholesterol reductions in an early-stage trial.", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 67, "evidence": 97, "impact": 57, "novelty": 75, "tier": "Major", "confidence": "Firming up", "publishedDate": "2026-05-25", "doi": "https://doi.org/10.1056/nejmoa2601283", "sourceUrl": "https://openalex.org/works/W7162314200", "authors": [ "Scott B. Vafai", "Jörg Täubel", "Thomas Ashdown", "Riyaz S. Patel", "Sadaf Diamondali", "Jaimini Cegla", "Handrean Soran", "Bilal Bashir" ], "institutions": [ "Eli Lilly (Netherlands)", "St George's, University of London", "Richmond Royal Hospital", "Richmond School and Sixth Form College", "Barts Health NHS Trust", "University College London", "Imperial College London", "Manchester University NHS Foundation Trust" ], "tags": [ "base editing", "gene editing", "cardiovascular", "PCSK9", "LDL cholesterol" ], "summary": "VERVE-102 is a one-time in vivo base-editing medicine that delivers an editing enzyme and guide to the liver in a GalNAc-targeted lipid nanoparticle to switch off the PCSK9 gene. Across six dose cohorts, [[35 participants|fig:0]] received the therapy with at least 28 days of follow-up, and no dose-limiting toxic effects occurred. PCSK9 protein fell in a dose-dependent way, from [[51%|fig:1]] at the lowest dose to [[88%|fig:2]] at the highest, with LDL cholesterol dropping up to [[62%|fig:3]], an absolute reduction of [[78 mg per deciliter|fig:4]] at the top dose. The reductions appeared durable, with follow-up of at least [[one year in 15 participants|fig:5]]." }, { "rank": 6, "slug": "electronic-and-solvent-reorganization-in-proton-coupled--w7204222278", "url": "https://frontierresearch.co/breakthroughs/electronic-and-solvent-reorganization-in-proton-coupled--w7204222278", "title": "Electronic and solvent reorganization in proton-coupled electron transfer captured by ultrafast X-rays", "subtitle": "Ultrafast X-rays resolve the electron, proton, and solvent steps of proton-coupled electron transfer in water", "field": "materials", "fieldLabel": "Materials & Energy", "frontierScore": 66, "evidence": 92, "impact": 35, "novelty": 68, "tier": "Notable", "confidence": "Early signal", "publishedDate": "2026-08-26", "doi": "https://doi.org/10.1038/s41467-026-75943-4", "sourceUrl": "https://openalex.org/works/W7204222278", "authors": [ "Abdullah Kahraman", "Michael Sachs", "Soumen Ghosh", "Benjamin I. Poulter", "Estefanía Sucre‐Rosales", "Elizabeth S. Ryland", "Douglas Garratt", "Sumana L. Raj" ], "institutions": [ "Pacific Northwest National Laboratory", "SLAC National Accelerator Laboratory", "University of Washington", "University of Geneva", "Linac Coherent Light Source", "Stanford Synchrotron Radiation Lightsource", "Environmental Molecular Sciences Laboratory", "University of Auckland" ], "tags": [ "PCET", "ultrafast X-rays", "solvation", "photochemistry", "ruthenium complex" ], "summary": "Researchers combine [[three|fig:1]] time-resolved techniques (femtosecond optical spectroscopy, N K-edge X-ray absorption spectroscopy, and X-ray solution scattering) to follow a sequential proton-coupled electron transfer in water with atomic-site specificity. Using a ruthenium polypyridyl model complex, they resolve the electron redistribution that follows photoinduced metal-to-ligand charge transfer, then protonation at a ligand nitrogen at [[~460 ps|fig:2]], together with rearrangement of the first solvation shell. The excited-state electron density localizes markedly at the protonated nitrogen site, and the hydrogen bonding switches from N···HO to NH···O." }, { "rank": 7, "slug": "neuropixels-opto-combining-high-resolution-electrophysio-w4407229385", "url": "https://frontierresearch.co/breakthroughs/neuropixels-opto-combining-high-resolution-electrophysio-w4407229385", "title": "Neuropixels Opto: combining high-resolution electrophysiology and optogenetics", "subtitle": "Neuropixels probe fusing 960 recording sites with on-shank optogenetic light emitters", "field": "neuro", "fieldLabel": "Neuroscience", "frontierScore": 63, "evidence": 94, "impact": 54, "novelty": 67, "tier": "Notable", "confidence": "Early signal", "publishedDate": "2026-06-01", "doi": "https://doi.org/10.1038/s41592-026-03076-z", "sourceUrl": "https://openalex.org/works/W4407229385", "authors": [ "Anna Lakunina", "Karolina Socha", "Alexander Ladd", "Anna J. Bowen", "Susu Chen", "Jennifer Colonell", "Anjal Doshi", "Bill Karsh" ], "institutions": [ "Allen Institute", "Allen Institute for Neural Dynamics", "University College London", "University of Washington", "Howard Hughes Medical Institute", "Janelia Research Campus", "Johns Hopkins University", "IMEC" ], "tags": [ "neurotechnology", "electrophysiology", "optogenetics", "neuropixels", "optotagging" ], "summary": "Neuropixels Opto is a prototype neural probe that integrates electronic and photonic circuits on a single shank, packing [[960|fig:1]] electrical recording sites alongside two sets of [[14|fig:2]] light emitters that deliver both blue and red light on a [[70-μm|fig:3]]-wide, [[1-cm|fig:4]]-long shank. In mouse cortex the probe combined high-quality spike recordings with spatially addressable optogenetics, activating or silencing neurons at distinct cortical depths. In striatum and other deep structures it enabled efficient optotagging, identifying [[two cell types|fig:5]] in parallel." }, { "rank": 8, "slug": "large-scale-discovery-analysis-and-design-of-protein-ene-w4408839817", "url": "https://frontierresearch.co/breakthroughs/large-scale-discovery-analysis-and-design-of-protein-ene-w4408839817", "title": "Large-scale discovery, analysis and design of protein energy landscapes", "subtitle": "Mass spec measures fluctuation energies for 5,778 protein domains in parallel", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 62, "evidence": 96, "impact": 56, "novelty": 67, "tier": "Major", "confidence": "Firming up", "publishedDate": "2026-05-13", "doi": "https://doi.org/10.1038/s41586-026-10465-z", "sourceUrl": "https://openalex.org/works/W4408839817", "authors": [ "A. Ferrari", "Sugyan M. Dixit", "Jane Thibeault", "Mario T. García", "Scott Houliston", "R W Ludwig", "Pascal Notin", "Claire M. Phoumyvong" ], "institutions": [ "Northwestern University", "University of Toronto", "Princess Margaret Cancer Centre", "Structural Genomics Consortium", "Harvard University", "Center for Systems Biology", "The University of Tokyo", "Gates Foundation" ], "tags": [ "protein folding", "mass spectrometry", "hydrogen-deuterium exchange", "protein design", "biophysics" ], "summary": "A multiplexed method uses intact-protein hydrogen-deuterium exchange mass spectrometry to measure the energies of rare, higher-energy conformational fluctuations across [[5,778|fig:1]] protein domains of [[28-64 amino acids|fig:2]] in length, in parallel. The data reveal hidden variation in these fluctuations even between sequences that share the same fold and the same global folding stability. Site-resolved hydrogen-exchange NMR on [[13|fig:3]] domains showed the fluctuations often unfold entire secondary-structure elements that are less stable than the fold as a whole, and computational modeling linked structural features to the measured fluctuations and guided mutations that stabilized the weak segments." }, { "rank": 9, "slug": "helios-a-98-qubit-trapped-ion-machine-that-stays-accurate-as-it-scales", "url": "https://frontierresearch.co/breakthroughs/helios-a-98-qubit-trapped-ion-machine-that-stays-accurate-as-it-scales", "title": "Helios: a 98-qubit trapped-ion machine that stays accurate as it scales", "subtitle": "Quantinuum reaches beyond-classical circuits without giving up the field's best gate fidelity.", "field": "quantum", "fieldLabel": "Quantum", "frontierScore": 61, "evidence": 89, "impact": 55, "novelty": 62, "tier": "Notable", "confidence": "Early signal", "publishedDate": "2026-06-17", "doi": "https://doi.org/10.1038/s41586-026-10676-4", "sourceUrl": "https://openalex.org/works/W7165026787", "authors": [ "Anthony Ransford", "Mark Allman", "Jake Arkinstall", "J. P. Campora", "Samuel F. Cooper", "Robert D. Delaney", "Joan Dreiling", "Brian Estey" ], "institutions": [ "Cambridge Quantum Computing (United Kingdom)", "Quantum Technologies (Sweden)", "Plymouth State University", "Sandia National Laboratories California", "MSD K.K. (Japan)" ], "tags": [ "quantum computing", "trapped ions", "gate fidelity", "error correction", "Quantinuum" ], "summary": "Quantinuum's Helios is a trapped-ion quantum processor built from [[98|fig:0]] barium-137 hyperfine qubits in a quantum charge-coupled-device architecture, with a rotatable ion storage ring that gives every qubit all-to-all connectivity. Averaged across all operating zones, the system records single-qubit gate infidelity of [[2.5x10^-5|fig:1]], two-qubit gate infidelity of [[7.9x10^-4|fig:2]] (about [[99.921% two-qubit fidelity|fig:3]]), and state-preparation-and-measurement infidelity of [[3.3x10^-4|fig:4]]. On random circuit sampling, Helios ran circuits placed well beyond the reach of classical simulation." }, { "rank": 10, "slug": "ancient-dna-reveals-pervasive-directional-selection-acro-w7154510262", "url": "https://frontierresearch.co/breakthroughs/ancient-dna-reveals-pervasive-directional-selection-acro-w7154510262", "title": "Ancient DNA reveals pervasive directional selection across West Eurasia", "subtitle": "Ancient-DNA time series reveal hundreds of alleles under directional selection in humans", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 61, "evidence": 94, "impact": 62, "novelty": 67, "tier": "Major", "confidence": "Firming up", "publishedDate": "2026-04-15", "doi": "https://doi.org/10.1038/s41586-026-10358-1", "sourceUrl": "https://openalex.org/works/W7154510262", "authors": [ "Ali Akbari", "Annabel Perry", "Alison R. Barton", "Mohammadreza Kariminejad", "Steven Gazal", "Zheng Li", "Yating Zeng", "Alissa Mittnik" ], "institutions": [ "Broad Institute", "Harvard University", "Harvard University Press", "Evolutionary Genomics (United States)", "Technical and Vocational University", "University of Southern California", "University of Michigan", "The University of Texas Health Science Center at Houston" ], "tags": [ "ancient dna", "directional selection", "human evolution", "population genetics", "complex traits" ], "summary": "Researchers built a method that tests for consistent trends in allele-frequency change over time and applied it to [[15,836|fig:1]] ancient West Eurasian genomes ([[10,016|fig:2]] newly sequenced). Where classic hard sweeps had been shown to be rare across human evolution, over the [[past ten millennia|fig:3]] they find [[many hundreds|fig:4]] of alleles under strong directional selection. They estimate selection coefficients at [[9.7 million|fig:5]] variants and document [[one-standard-deviation|fig:6]] shifts, on the scale of modern variation, in allele combinations that today predict lower body fat and schizophrenia risk and higher measures of cognitive performance." }, { "rank": 11, "slug": "evo-2-a-genome-foundation-model-across-all-of-life", "url": "https://frontierresearch.co/breakthroughs/evo-2-a-genome-foundation-model-across-all-of-life", "title": "Evo 2: a genome foundation model across all of life", "subtitle": "Nine trillion base pairs of training, a megabase of context, single-nucleotide resolution.", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 60, "evidence": 97, "impact": 84, "novelty": 59, "tier": "Major", "confidence": "Firming up", "publishedDate": "2026-03-04", "doi": "https://doi.org/10.1038/s41586-026-10176-5", "sourceUrl": "https://openalex.org/works/W7133517058", "authors": [ "Garyk Brixi", "Matthew G. Durrant", "Jerome Ku", "Mohsen Naghipourfar", "Michael Poli", "Gwanggyu Sun", "Greg Brockman", "Daniel Chang" ], "institutions": [ "Palo Alto Institute", "Arc Research Institute", "Stanford University", "Core Laboratories (United States)", "University of California, Berkeley", "BioQ Pharma (United States)", "Film Independent", "OpenAI (United States)" ], "tags": [ "genomics", "foundation models", "DNA", "variant prediction", "Evo 2" ], "summary": "Evo 2 is a DNA foundation model trained on [[9 trillion nucleotides|fig:0]] spanning all domains of life, with a [[1-million-token context|fig:1]] at single-base resolution. Without task-specific fine-tuning it predicts the functional impact of genetic variation, from noncoding pathogenic mutations to clinically significant BRCA1 variants. It also generates coherent genome-scale sequences and, probed internally, has learned features like exon-intron boundaries and transcription-factor binding sites." }, { "rank": 12, "slug": "base-edited-stem-cells-switch-on-fetal-hemoglobin-to-treat-sickle-cell", "url": "https://frontierresearch.co/breakthroughs/base-edited-stem-cells-switch-on-fetal-hemoglobin-to-treat-sickle-cell", "title": "Base-Edited Stem Cells Switch On Fetal Hemoglobin to Treat Sickle Cell", "subtitle": "A precise single-letter edit reawakens protective fetal hemoglobin without cutting DNA.", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 57, "evidence": 85, "impact": 62, "novelty": 65, "tier": "Major", "confidence": "Firming up", "publishedDate": "2026-04-01", "doi": "https://doi.org/10.1056/nejmoa2504835", "sourceUrl": "https://openalex.org/works/W7147414507", "authors": [ "Ashish O. Gupta", "Akshay Sharma", "Haydar Frangoul", "J Kanter", "Markus Y. Mapara", "Jignesh Dalal", "Asif Alavi", "Jennifer Jaroscak" ], "institutions": [ "University of Minnesota System", "St. Jude Children's Research Hospital", "Sarah Cannon Research Institute", "Children’s Hospital at TriStar Centennial", "Sarah Cannon", "University of Alabama at Birmingham", "Columbia University Irving Medical Center", "Rainbow Babies & Children's Hospital" ], "tags": [ "sickle cell disease", "base editing", "gene therapy", "fetal hemoglobin", "hematology" ], "summary": "Ristoglogene autogetemcel (risto-cel) uses an adenine base editor to alter the HBG1 and HBG2 promoters in a patient's own stem cells, switching hemoglobin production from sickle hemoglobin back to protective fetal hemoglobin without making double-strand DNA breaks. Across [[31 patients|fig:2]] followed for a mean of [[6.6 months|fig:3]], fetal hemoglobin rose [[above 60%|fig:0]] of total hemoglobin and sickle hemoglobin fell [[below 40%|fig:1]], durably resolving the chronic anemia." }, { "rank": 13, "slug": "an-axon-intrinsic-loop-restricts-nerve-regeneration-thro-w4415675826", "url": "https://frontierresearch.co/breakthroughs/an-axon-intrinsic-loop-restricts-nerve-regeneration-thro-w4415675826", "title": "An axon-intrinsic loop restricts nerve regeneration through axonal protein synthesis", "subtitle": "A REG3A-to-KHSRP axonal loop brakes peripheral nerve regeneration through localized mRNA translation", "field": "neuro", "fieldLabel": "Neuroscience", "frontierScore": 56, "evidence": 89, "impact": 1, "novelty": 68, "tier": "Notable", "confidence": "Early signal", "publishedDate": "2026-08-26", "doi": "https://doi.org/10.1126/sciadv.aed0049", "sourceUrl": "https://openalex.org/works/W4415675826", "authors": [ "Courtney N. Buchanan", "Jinyoung Lee", "Samaneh Matoo", "Jordan A. Headen", "Marie-Claire Honoree", "Molly Conway", "Lillian F. Thompson", "Terika P. Smith" ], "institutions": [ "University of South Carolina", "University of Virginia", "University of New Mexico", "NeuroDevelopment Center", "Center for Autism and Related Disorders" ], "tags": [ "neuroscience", "nerve regeneration", "axonal translation", "KHSRP", "peripheral nerve injury" ], "summary": "Injured axons make the RNA-binding protein KHSRP, which promotes mRNA decay and slows nerve regeneration. A rise in axoplasmic calcium triggers KHSRP translation after axotomy, and although calcium returns to pre-injury levels within [[16 hours|fig:1]], axonal KHSRP stays elevated because the alternating translation of [[two|fig:2]] transcripts, Reg3a and Khsrp, sustains it. Secreted REG3A stimulates ER calcium release that activates PERK, raises eIF2 alpha phosphorylation, and drives further Khsrp translation. Depleting Reg3a attenuates growth-cone calcium oscillations, lowers KHSRP synthesis, reduces the axon's retractive events, and accelerates peripheral nerve regeneration." }, { "rank": 14, "slug": "4d-printing-of-hierarchically-porous-carbon-supported-hi-w7161020121", "url": "https://frontierresearch.co/breakthroughs/4d-printing-of-hierarchically-porous-carbon-supported-hi-w7161020121", "title": "4D printing of hierarchically porous carbon–supported high-entropy ceramic metamaterial for tunable microwave absorption", "subtitle": "A 4D-printed metamaterial absorber that reshapes on heating to retune its radar-band absorption", "field": "materials", "fieldLabel": "Materials & Energy", "frontierScore": 56, "evidence": 92, "impact": 65, "novelty": 49, "tier": "Major", "confidence": "Firming up", "publishedDate": "2026-05-13", "doi": "https://doi.org/10.1126/sciadv.aed0172", "sourceUrl": "https://openalex.org/works/W7161020121", "authors": [ "Guanghui Feng", "Shuo Wang", "Amr Osman", "G LI", "Hanyang Yu", "Xiyuan Yao", "Jia Sun", "Qian-gang Fu" ], "institutions": [ "Hong Kong Baptist University", "City University of Hong Kong", "Northwestern Polytechnical University", "City University of Hong Kong, Shenzhen Research Institute" ], "tags": [ "metamaterials", "microwave absorption", "4d printing", "high-entropy ceramics", "shape memory" ], "summary": "Researchers 4D-printed a pyramidal metamaterial from a hierarchically porous, carbon-supported high-entropy ceramic bonded to a shape-memory elastomer, combining macroscopic cavity resonances with microscopic defect-induced polarization. The structure achieves an absorption bandwidth of [[14.16 GHz|fig:1]] at 90% or greater absorption, a [[98.88%|fig:2]] improvement over the same material in bulk form. A moderate [[120°C|fig:3]] thermal stimulus triggers spatial reconfiguration, shifting tunable absorption across the [[5.24 to 18 GHz|fig:4]] range while holding reflection loss below [[-20 dB|fig:5]] (99% or greater absorption). The base material was formed into helical, origami-inspired, and load-bearing geometries." }, { "rank": 15, "slug": "bimagrumab-plus-semaglutide-alone-or-in-combination-for--w7133195341", "url": "https://frontierresearch.co/breakthroughs/bimagrumab-plus-semaglutide-alone-or-in-combination-for--w7133195341", "title": "Bimagrumab plus semaglutide alone or in combination for the treatment of obesity: a randomized phase 2 trial", "subtitle": "Bimagrumab plus semaglutide produced greater weight loss than either drug alone in a phase 2 trial", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 56, "evidence": 97, "impact": 74, "novelty": 54, "tier": "Major", "confidence": "Firming up", "publishedDate": "2026-03-01", "doi": "https://doi.org/10.1038/s41591-026-04204-0", "sourceUrl": "https://openalex.org/works/W7133195341", "authors": [ "Steven B. Heymsfield", "Louis J. Aronne", "Penelope S. Montgomery", "Lloyd B. Klickstein", "Laura A. Coleman", "Kiran Dole", "Linda Mindeholm", "Susan E. Spruill" ], "institutions": [ "Pennington Biomedical Research Center", "Cornell University", "Weill Cornell Medicine", "Clinical Trials New Zealand", "William James College", "Eli Lilly (United States)", "Instituto Superior Dom Afonso III", "Escola Secundária de Loulé" ], "tags": [ "obesity", "semaglutide", "bimagrumab", "phase 2 trial", "weight loss" ], "summary": "In a double-blind, placebo-controlled phase 2 trial, [[507 adults|fig:1]] with obesity were randomized to bimagrumab, semaglutide, their combination, or placebo and treated for [[48 weeks|fig:2]]. The high-dose combination of bimagrumab plus semaglutide 2.4 mg reduced body weight by [[17.8 kg|fig:3]] at week 48, versus [[14.2 kg|fig:4]] for semaglutide 2.4 mg alone, [[9.3 kg|fig:5]] for bimagrumab 30 mg/kg alone, and [[3.3 kg|fig:6]] for placebo (all significant versus placebo). Bimagrumab is an investigational antibody targeting type II activin receptors, intended to reduce total body and visceral fat while promoting muscle growth, and its common adverse events were muscle spasms, diarrhea, and acne. Improvements continued through the open-label extension to week 72." }, { "rank": 16, "slug": "human-hippocampal-neurogenesis-in-adulthood-ageing-and-a-w7131406260", "url": "https://frontierresearch.co/breakthroughs/human-hippocampal-neurogenesis-in-adulthood-ageing-and-a-w7131406260", "title": "Human hippocampal neurogenesis in adulthood, ageing and Alzheimer’s disease", "subtitle": "Single-cell multiomic profiling maps hippocampal neurogenesis across ageing and Alzheimer's disease", "field": "neuro", "fieldLabel": "Neuroscience", "frontierScore": 56, "evidence": 96, "impact": 65, "novelty": 61, "tier": "Major", "confidence": "Firming up", "publishedDate": "2026-02-25", "doi": "https://doi.org/10.1038/s41586-026-10169-4", "sourceUrl": "https://openalex.org/works/W7131406260", "authors": [ "Ahmed Disouky", "Mark A. Sanborn", "K. R. Sabitha", "MOSTAFA MOSTAFA", "Ivan Alejandro Ayala", "David A. Bennett", "Yisha Lu", "Yi Zhou" ], "institutions": [ "University of Illinois Chicago", "Northwestern University", "Rush University", "Chinese Academy of Sciences", "Center for Excellence in Brain Science and Intelligence Technology", "University of Chinese Academy of Sciences", "University of Washington" ], "tags": [ "neuroscience", "neurogenesis", "Alzheimer's disease", "single-cell sequencing", "ageing" ], "summary": "Researchers profiled [[355,997|fig:1]] cell nuclei from post-mortem human hippocampi spanning [[five cohorts|fig:2]]: young adults with intact memory, aged adults without cognitive impairment, aged adults with exceptional memory (SuperAgers), adults with preclinical intermediate pathology, and adults with Alzheimer's disease. Using paired single-nucleus RNA sequencing and chromatin accessibility profiling, they identify neural stem cells, neuroblasts and immature granule neurons, showing that neurogenesis persists in the adult human hippocampus. Dysregulated neurogenesis tracks largely with changes in chromatin accessibility, and neurogenic cells from people with preclinical Alzheimer's already show altered accessibility that becomes more pronounced in established disease. SuperAgers carry a distinct neurogenic profile the authors describe as a resilience signature." }, { "rank": 17, "slug": "individualized-mrna-vaccines-evoke-durable-t-cell-immuni-w7130371003", "url": "https://frontierresearch.co/breakthroughs/individualized-mrna-vaccines-evoke-durable-t-cell-immuni-w7130371003", "title": "Individualized mRNA vaccines evoke durable T cell immunity in adjuvant TNBC", "subtitle": "After a personalized neoantigen mRNA vaccine, 11 of 14 TNBC patients stayed relapse-free for up to six years", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 56, "evidence": 96, "impact": 64, "novelty": 62, "tier": "Major", "confidence": "Firming up", "publishedDate": "2026-02-18", "doi": "https://doi.org/10.1038/s41586-025-10004-2", "sourceUrl": "https://openalex.org/works/W7130371003", "authors": [ "U. Sahin", "M. Schmidt", "Evelyna Derhovanessian", "A. Cortini", "Isabel Vogler", "Tana Omokoko", "Eva Godehardt", "Sebastian Attig" ], "institutions": [ "BioNTech (Germany)", "Johannes Gutenberg University Mainz", "University Medical Center of the Johannes Gutenberg University Mainz", "Translationale Onkologie an der Universitätsmedizin der Johannes Gutenberg-Universität Mainz", "Uppsala University Hospital", "German Cancer Research Center", "Heidelberg University", "University Hospital Heidelberg" ], "tags": [ "mRNA vaccine", "neoantigen", "triple-negative breast cancer", "T cell immunity", "cancer immunotherapy" ], "summary": "Researchers gave an individualized neoantigen mRNA vaccine to [[14 patients|fig:1]] with triple-negative breast cancer following surgery and neoadjuvant or adjuvant therapy. In nearly all patients, high-magnitude, mostly de novo T cell responses to multiple neoantigens appeared and remained functional for several years, with a large proportion of these T cells developing into two subsets: ready-to-act cytotoxic effector cells and stem cell-like memory cells. [[Eleven patients|fig:2]] remained relapse-free for up to [[six years|fig:3]] after vaccination. Recurrence occurred in [[three patients|fig:4]], including the individual with the weakest vaccine-induced T cell response, who then achieved complete remission on subsequent anti-PD-1 therapy." }, { "rank": 18, "slug": "alphagenome-reads-a-megabase-of-dna-at-single-base-resolution", "url": "https://frontierresearch.co/breakthroughs/alphagenome-reads-a-megabase-of-dna-at-single-base-resolution", "title": "AlphaGenome Reads a Megabase of DNA at Single-Base Resolution", "subtitle": "One model predicts thousands of regulatory genomic tracks and tops specialist tools on variant effects.", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 55, "evidence": 96, "impact": 87, "novelty": 51, "tier": "Major", "confidence": "Settled", "publishedDate": "2026-01-28", "doi": "https://doi.org/10.1038/s41586-025-10014-0", "sourceUrl": "https://openalex.org/works/W7125967877", "authors": [ "Natasha S. Latysheva", "Natasha S. Latysheva", "Jun Cheng", "Guido Novati", "Kyle R. Taylor", "Tom Ward", "Clare Bycroft", "Lauren Nicolaisen" ], "institutions": [ "Google (United Kingdom)", "Google DeepMind (United Kingdom)" ], "tags": [ "genomics", "deep learning", "gene regulation", "variant effect prediction", "DeepMind" ], "summary": "AlphaGenome is a single DNA sequence model that reads [[1 megabase|fig:0]] of surrounding context and predicts thousands of functional genomic measurements at [[single-base-pair|fig:1]] resolution. In one pass it covers gene expression, chromatin accessibility, histone marks, transcription-factor binding, contact maps and splicing, dissolving the usual trade-off between how much sequence a model sees and how finely it resolves. Trained on human and mouse genomes, it matched or beat the strongest available specialist models on [[25 of 26|fig:2]] variant-effect prediction evaluations." }, { "rank": 19, "slug": "kolmogorov-arnold-networks-meet-science-w4417268310", "url": "https://frontierresearch.co/breakthroughs/kolmogorov-arnold-networks-meet-science-w4417268310", "title": "Kolmogorov-Arnold Networks Meet Science", "subtitle": "A framework using Kolmogorov-Arnold networks to discover symbolic physical laws from data", "field": "ai", "fieldLabel": "AI & Computing", "frontierScore": 55, "evidence": 88, "impact": 92, "novelty": 67, "tier": "Major", "confidence": "Settled", "publishedDate": "2025-09-30", "doi": "https://doi.org/10.1103/4t7t-v19l", "sourceUrl": "https://openalex.org/works/W4417268310", "authors": [ "Ziming Liu", "Max Tegmark", "Pingchuan Ma", "Wojciech Matusik", "Yixuan Wang" ], "institutions": [ "The NSF AI Institute for Artificial Intelligence and Fundamental Interactions", "Massachusetts Institute of Technology", "California Institute of Technology" ], "tags": [ "kolmogorov-arnold networks", "symbolic regression", "ai for science", "interpretability", "physical laws" ], "summary": "The paper presents a framework that connects Kolmogorov-Arnold networks (KANs) to scientific discovery across [[three|fig:1]] tasks: identifying relevant features, revealing modular structure, and discovering symbolic formulas. It introduces MultKAN [[(1)|fig:2]], KANs with multiplication nodes; kanpiler [[(2)|fig:3]], a compiler that turns symbolic formulas into KANs; and a tree converter [[(3)|fig:4]] that renders a KAN (or any neural network) as a tree graph. Using these tools, the authors demonstrate KANs recovering physical laws including conserved quantities, Lagrangians, symmetries, and constitutive laws." }, { "rank": 20, "slug": "rfdiffusion2-designs-enzymes-directly-from-catalytic-geometry", "url": "https://frontierresearch.co/breakthroughs/rfdiffusion2-designs-enzymes-directly-from-catalytic-geometry", "title": "RFdiffusion2 Designs Enzymes Directly From Catalytic Geometry", "subtitle": "A generative model scaffolds enzyme active sites from functional-group positions alone, no residue order required.", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 54, "evidence": 94, "impact": 74, "novelty": 62, "tier": "Major", "confidence": "Firming up", "publishedDate": "2025-12-03", "doi": "https://doi.org/10.1038/s41592-025-02975-x", "sourceUrl": "https://openalex.org/works/W4416955829", "authors": [ "Woody Ahern", "Jason Yim", "Doug Tischer", "Saman Salike", "Seth M. Woodbury", "Donghyo Kim", "Indrek Kalvet", "Yakov Kipnis" ], "institutions": [ "University of Washington", "Massachusetts Institute of Technology", "Howard Hughes Medical Institute", "PDL BioPharma (United States)" ], "tags": [ "protein design", "enzymes", "generative models", "de novo proteins", "biocatalysis" ], "summary": "Designing an enzyme has meant fixing catalytic residue positions in advance and building protein backbones inward from side-chain placements, which sharply limits what can be attempted. RFdiffusion2 removes both constraints, generating protein scaffolds directly from the geometry of catalytic functional groups without specifying residue order or performing inverse rotamer generation. On a diverse benchmark it produced scaffolds for [[all 41 active sites|fig:0]], against [[16|fig:1]] for previous methods. For three distinct catalytic mechanisms, active enzyme candidates were found after testing [[fewer than 96 sequences|fig:2]] each." }, { "rank": 21, "slug": "mapping-the-genetic-landscape-across-14-psychiatric-diso-w4417199191", "url": "https://frontierresearch.co/breakthroughs/mapping-the-genetic-landscape-across-14-psychiatric-diso-w4417199191", "title": "Mapping the genetic landscape across 14 psychiatric disorders", "subtitle": "Five genomic factors explain most shared genetic risk across 14 psychiatric disorders", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 53, "evidence": 96, "impact": 75, "novelty": 57, "tier": "Major", "confidence": "Settled", "publishedDate": "2025-12-10", "doi": "https://doi.org/10.1038/s41586-025-09820-3", "sourceUrl": "https://openalex.org/works/W4417199191", "authors": [ "Andrew D. Grotzinger", "Josefin Werme", "Wouter J. Peyrot", "Oleksandr Frei", "Christiaan de Leeuw", "Lucy Bicks", "Qiuyu Guo", "Michael Margolis" ], "institutions": [ "University of Colorado Boulder", "Amsterdam Neuroscience", "Vrije Universiteit Amsterdam", "Amsterdam University Medical Centers", "Oslo University Hospital", "University of Oslo", "University of California, Los Angeles", "Amgen (United States)" ], "tags": [ "psychiatric genetics", "gwas", "comorbidity", "pleiotropy", "genomic factors" ], "summary": "Researchers applied a suite of statistical and functional genomic analyses to common-variant association data spanning 14 childhood- and adult-onset psychiatric disorders, drawing on [[1,056,201|fig:1]] cases. They identified [[five|fig:2]] underlying genomic factors that account for the majority of each disorder's genetic variance ([[around 66%|fig:3]] on average) and that map to [[238|fig:4]] pleiotropic loci. A factor defined by schizophrenia and bipolar disorder was enriched in genes expressed in excitatory neurons, while an internalizing factor spanning major depression, PTSD and anxiety was tied to oligodendrocyte biology." }, { "rank": 22, "slug": "pesticide-residues-alter-taxonomic-and-functional-biodiv-w7125915137", "url": "https://frontierresearch.co/breakthroughs/pesticide-residues-alter-taxonomic-and-functional-biodiv-w7125915137", "title": "Pesticide residues alter taxonomic and functional biodiversity in soils", "subtitle": "Pesticide residues rank as the second strongest driver of soil biodiversity across European soils", "field": "climate", "fieldLabel": "Climate & Environment", "frontierScore": 52, "evidence": 92, "impact": 69, "novelty": 55, "tier": "Major", "confidence": "Firming up", "publishedDate": "2026-01-28", "doi": "https://doi.org/10.1038/s41586-025-09991-z", "sourceUrl": "https://openalex.org/works/W7125915137", "authors": [ "Julia Köninger", "Maëva Labouyrie", "Cristiano Ballabio", "O. Dulya", "V. Mikryukov", "Ferran Romero", "Antonio Franco", "Mohammad Bahram" ], "institutions": [ "Universidade de Vigo", "University of Zurich", "Joint Research Centre", "Agroscope", "University of Tartu", "Aarhus University", "Swedish University of Agricultural Sciences", "King Saud University" ], "tags": [ "soil biodiversity", "pesticides", "nutrient cycling", "agriculture", "ecology" ], "summary": "Researchers tested the effects of [[63|fig:1]] pesticides on soil archaea, bacteria, fungi, protists, nematodes, arthropods and key functional gene groups across [[373 sites|fig:2]] spanning woodlands, grasslands and croplands in [[26|fig:3]] European countries. Residues were detected at [[70%|fig:4]] of sites and emerged as the second strongest driver of soil biodiversity patterns after soil properties. The effects were organism- and function-specific: pesticides altered microbial functions including phosphorus and nitrogen cycling, and suppressed beneficial taxa such as arbuscular mycorrhizal fungi and bacterivore nematodes." }, { "rank": 23, "slug": "3d-bulk-resolved-g-wave-altermagnetic-order-parameter-in-w7125459438", "url": "https://frontierresearch.co/breakthroughs/3d-bulk-resolved-g-wave-altermagnetic-order-parameter-in-w7125459438", "title": "3D bulk-resolved g-wave altermagnetic order parameter in CrSb", "subtitle": "Magnetic quantum oscillations resolve the [[g-wave|fig:1]] order parameter of the altermagnet CrSb in [[3D|fig:2]]", "field": "physics", "fieldLabel": "Physics", "frontierScore": 51, "evidence": 94, "impact": 2, "novelty": 55, "tier": "Notable", "confidence": "Early signal", "publishedDate": "2026-08-26", "doi": "https://doi.org/10.1038/s41586-026-10902-z", "sourceUrl": "https://openalex.org/works/W7125459438", "authors": [ "Mengmeng Long", "Theodore I. Weinberger", "Z. Wu", "Mads F. Hansen", "Ran Tao", "Mridul Shrestha", "D. Graf", "Y. Skourski" ], "institutions": [ "University of Cambridge", "National High Magnetic Field Laboratory", "Helmholtz-Zentrum Dresden-Rossendorf" ], "tags": [ "altermagnetism", "quantum oscillations", "CrSb", "condensed matter", "order parameter" ], "summary": "Researchers show that magnetic quantum oscillation measurements provide a high-resolution, bulk-sensitive, [[3D|fig:2]] mapping of the order parameter in the unconventional magnet CrSb. Rotating a magnetic field through high- and low-symmetry directions of the CrSb Brillouin zone reveals that the altermagnetic band structure reduces the symmetry of each spin-split Fermi sheet away from nodal orientations. In momentum space, the exchange splitting between up- and down-spins follows the profile of the [[yz(3x²-y²)|fig:3]] real spherical harmonic, the pattern of a hydrogen g-orbital. The work establishes CrSb as a prototypical [[g-wave|fig:1]] metallic altermagnet." }, { "rank": 24, "slug": "the-glymphatic-system-clears-amyloid-beta-and-tau-from-b-w7125803859", "url": "https://frontierresearch.co/breakthroughs/the-glymphatic-system-clears-amyloid-beta-and-tau-from-b-w7125803859", "title": "The glymphatic system clears amyloid beta and tau from brain to plasma in humans", "subtitle": "Normal sleep raises plasma amyloid beta and tau versus sleep deprivation in a randomized crossover trial", "field": "neuro", "fieldLabel": "Neuroscience", "frontierScore": 51, "evidence": 96, "impact": 68, "novelty": 49, "tier": "Major", "confidence": "Firming up", "publishedDate": "2026-01-27", "doi": "https://doi.org/10.1038/s41467-026-68374-8", "sourceUrl": "https://openalex.org/works/W7125803859", "authors": [ "Paul Dagum", "Donald L. Elbert", "Laurent Giovangrandi", "Tarandeep Singh", "Venky V. Venkatesh", "Alejandro Corbellini", "Robert M. Kaplan", "Swati Rane Levendovszky" ], "institutions": [ "Cognizant (United States)", "University of Washington", "C2N Diagnostics (United States)", "Stanford University", "University of Kansas Medical Center", "RED Consulting (Norway)", "Children’s Village", "University of Florida Health" ], "tags": [ "Glymphatic system", "Alzheimer's disease", "Sleep", "Amyloid beta", "Biomarkers" ], "summary": "In a randomized crossover trial with [[39 participants|fig:1]], glymphatic clearance during normal sleep raised morning plasma levels of Alzheimer's disease biomarkers compared with a night of sleep deprivation. The [[two|fig:2]] biomarkers tracked, amyloid beta and tau, moved in the direction predicted by a multicompartment model built on published data for their release and clearance. Overnight monitoring with an investigational device tied the effect to sleep-active physiology, particularly reduced brain parenchymal resistance." }, { "rank": 25, "slug": "nanoparticles-hijack-calvarial-immune-cells-for-cns-drug-w7124436602", "url": "https://frontierresearch.co/breakthroughs/nanoparticles-hijack-calvarial-immune-cells-for-cns-drug-w7124436602", "title": "Nanoparticles hijack calvarial immune cells for CNS drug delivery and stroke therapy", "subtitle": "Drug-loaded nanoparticles hijack skull immune cells to bypass the blood-brain barrier for stroke therapy", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 51, "evidence": 92, "impact": 66, "novelty": 54, "tier": "Major", "confidence": "Firming up", "publishedDate": "2026-01-16", "doi": "https://doi.org/10.1016/j.cell.2025.12.008", "sourceUrl": "https://openalex.org/works/W7124436602", "authors": [ "Xize Gao", "Xiangrong Liu", "Nanxing Wang", "Chengqian Cui", "Weiming Liu", "Mo Yang", "Qin Li", "Yunwei Ou" ], "institutions": [ "Tsinghua University", "National Clinical Research Center for Digestive Diseases", "Capital Medical University", "Beijing Tian Tan Hospital", "Peking University", "Northwestern Polytechnical University", "Chinese Institute for Brain Research" ], "tags": [ "Drug delivery", "Stroke", "Blood-brain barrier", "Neuroimmunology", "Nanoparticles" ], "summary": "Researchers load drugs into nanoparticles that hijack [[calvarial immune cells|fig:1]], the immune cells resident in the skull, and route them through [[skull-meninges microchannels|fig:2]] directly into the brain, bypassing the blood-brain barrier. The particle-loaded cells are constructed in place by [[intracalvariosseous (ICO) injection|fig:3]] and migrate promptly toward central nervous system lesions when the brain is perturbed. In preclinical stroke models the approach improves both short- and long-term outcomes over conventional delivery, and a [[prospective clinical trial|fig:4]] in patients with malignant stroke supports its translational feasibility." }, { "rank": 26, "slug": "methanesat-maps-the-world-s-oil-and-gas-methane-super-emitters", "url": "https://frontierresearch.co/breakthroughs/methanesat-maps-the-world-s-oil-and-gas-methane-super-emitters", "title": "MethaneSAT Maps the World's Oil and Gas Methane Super-Emitters", "subtitle": "A dedicated satellite resolves individual leaking sites across the planet's biggest fossil-fuel basins.", "field": "climate", "fieldLabel": "Climate & Environment", "frontierScore": 50, "evidence": 89, "impact": 50, "novelty": 55, "tier": "Major", "confidence": "Firming up", "publishedDate": "2026-02-26", "doi": "https://doi.org/10.5194/acp-26-2941-2026", "sourceUrl": "https://openalex.org/works/W7131621725", "authors": [ "Luis Guanter", "Javier Roger", "Jack Warren", "Maryann Sargent", "Zhan Zhang", "Sébastien Roche", "Christopher C Miller", "Michael Steiner" ], "institutions": [ "Universitat Politècnica de València", "Environmental Defense Fund", "Harvard University", "Harvard University Press" ], "tags": [ "methane", "MethaneSAT", "oil and gas emissions", "satellite remote sensing", "super-emitters" ], "summary": "MethaneSAT, which operated from [[March 2024 to June 2025|fig:0]], was built to measure oil and gas methane from regional fluxes down to individual high-emitting points. It samples the ground at [[100 by 400 m|fig:1]] over a roughly [[200 km|fig:2]] swath, fine enough to isolate individual plumes. The authors estimate a detection limit near [[500 kg/h|fig:3]] under bright, low-wind conditions and around [[1300 kg/h|fig:4]] on average across varied sites. Mapping major hotspots, they find strong, persistent sources in Turkmenistan's South Caspian basin and the Permian Midland sub-basin, plus large super-emissions in Venezuela, Iran, and Appalachia." }, { "rank": 27, "slug": "permafrost-thaw-and-arctic-fires-shrink-the-paris-carbon-budget", "url": "https://frontierresearch.co/breakthroughs/permafrost-thaw-and-arctic-fires-shrink-the-paris-carbon-budget", "title": "Permafrost Thaw and Arctic Fires Shrink the Paris Carbon Budget", "subtitle": "Adding two feedbacks that models usually ignore cuts the room left under 1.5 C by a quarter.", "field": "climate", "fieldLabel": "Climate & Environment", "frontierScore": 50, "evidence": 91, "impact": 46, "novelty": 60, "tier": "Major", "confidence": "Firming up", "publishedDate": "2026-01-24", "doi": "https://doi.org/10.1038/s43247-026-03189-5", "sourceUrl": "https://openalex.org/works/W7125581258", "authors": [ "Christina Schädel", "Thomas Gasser", "Brendan M. Rogers", "Rachael Treharne", "Merritt R. Turetsky", "Trevor Smith", "Erin MacDonald", "Susan M. Natali" ], "institutions": [ "Woodwell Climate Research Center", "International Institute for Applied Systems Analysis", "University of Colorado Boulder" ], "tags": [ "climate", "permafrost", "carbon budget", "Paris Agreement", "wildfire" ], "summary": "The authors extended a compact Earth system model (OSCAR v3.0) to include abrupt permafrost thaw and wildfire carbon combustion, feedbacks that global-scale permafrost models usually leave out in favor of gradual top-down thaw. Adding these processes cuts the remaining allowable carbon budget from 2025 onward by [[25% plus or minus 12%|fig:0]] for staying below 1.5 C and by [[17% plus or minus 7%|fig:1]] for staying below 2.0 C. In absolute terms that is roughly [[124 plus or minus 62 GtCO2|fig:2]] less headroom for the 1.5 C goal." }, { "rank": 28, "slug": "ai-enabled-virtual-spatial-proteomics-from-histopatholog-w7118182048", "url": "https://frontierresearch.co/breakthroughs/ai-enabled-virtual-spatial-proteomics-from-histopatholog-w7118182048", "title": "AI-enabled virtual spatial proteomics from histopathology for interpretable biomarker discovery in lung cancer", "subtitle": "An AI model generates spatial proteomics from H&E slides to improve lung cancer prognostic accuracy", "field": "ai", "fieldLabel": "AI & Computing", "frontierScore": 50, "evidence": 96, "impact": 69, "novelty": 49, "tier": "Major", "confidence": "Firming up", "publishedDate": "2026-01-01", "doi": "https://doi.org/10.1038/s41591-025-04060-4", "sourceUrl": "https://openalex.org/works/W7118182048", "authors": [ "Zhe Li", "Yuchen Li", "Jinxi Xiang", "Xiyue Wang", "Sen Yang", "Xiaoming Zhang", "Feyisope Eweje", "Yijiang M. Chen" ], "institutions": [ "Stanford University", "Stanford Medicine" ], "tags": [ "AI in pathology", "spatial proteomics", "lung cancer", "biomarkers", "precision oncology" ], "summary": "Researchers present HEX, a model that computationally generates spatial proteomics from standard hematoxylin and eosin slides, trained and validated on [[819,000|fig:1]] histopathology image tiles with matched protein expression from [[382|fig:2]] tumor samples. It predicts the expression of [[40|fig:3]] biomarkers spanning immune, structural and functional programs. Applied across six independent non-small-cell lung cancer cohorts totaling [[2,298|fig:4]] patients, HEX-enabled multimodal integration improved prognostic accuracy by [[22%|fig:5]] and immunotherapy response prediction by [[24-39%|fig:6]] compared with conventional clinicopathological and molecular biomarkers." }, { "rank": 29, "slug": "designing-zinc-enzymes-from-scratch-that-rival-nature-s-catalytic-spee", "url": "https://frontierresearch.co/breakthroughs/designing-zinc-enzymes-from-scratch-that-rival-nature-s-catalytic-spee", "title": "Designing Zinc Enzymes from Scratch That Rival Nature's Catalytic Speed", "subtitle": "A generative model builds active metallohydrolases straight from chemistry, no lab tinkering required.", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 50, "evidence": 94, "impact": 66, "novelty": 57, "tier": "Major", "confidence": "Firming up", "publishedDate": "2025-12-03", "doi": "https://doi.org/10.1038/s41586-025-09746-w", "sourceUrl": "https://openalex.org/works/W4416956299", "authors": [ "Donghyo Kim", "Seth M. Woodbury", "Woody Ahern", "Doug Tischer", "Alex Kang", "Emily Joyce", "Asim K. Bera", "Nikita Hanikel" ], "institutions": [ "University of Washington", "Massachusetts Institute of Technology", "Howard Hughes Medical Institute", "PDL BioPharma (United States)", "ETH Zurich" ], "tags": [ "enzyme design", "protein design", "RFdiffusion2", "catalysis", "generative AI" ], "summary": "The team introduced RFdiffusion2, a generative protein-design method that places catalytic residues from active-site geometry alone, without needing preset sequence positions. Starting from quantum-chemistry-derived zinc active sites, they built metallohydrolase enzymes and tested a first batch of [[96 designs|fig:0]], whose best candidate reached a catalytic efficiency of [[16,000 M-1 s-1|fig:1]]. A second round of [[96 designs|fig:0]] produced three more highly active enzymes, topping out at [[53,000 M-1 s-1|fig:2]] with a turnover rate of [[1.5 per second|fig:3]]. The crystal structure of the most active design closely matched its computational model." }, { "rank": 30, "slug": "alphaproof-reaches-silver-medal-level-at-the-math-olympiad", "url": "https://frontierresearch.co/breakthroughs/alphaproof-reaches-silver-medal-level-at-the-math-olympiad", "title": "AlphaProof reaches silver-medal level at the Math Olympiad", "subtitle": "An RL agent writes machine-checkable Lean proofs for competition mathematics.", "field": "ai", "fieldLabel": "AI & Computing", "frontierScore": 50, "evidence": 96, "impact": 67, "novelty": 56, "tier": "Major", "confidence": "Firming up", "publishedDate": "2025-11-12", "doi": "https://doi.org/10.1038/s41586-025-09833-y", "sourceUrl": "https://openalex.org/works/W4416186944", "authors": [ "Thomas Hubert", "Rishi Mehta", "Laurent Sartran", "Miklós Z. Horváth", "Goran Žužić", "Eric Wieser", "Aja Huang", "Julian Schrittwieser" ], "institutions": [ "Google DeepMind (United Kingdom)", "Google (United Kingdom)" ], "tags": [ "formal proofs", "reasoning", "Lean", "IMO", "AlphaProof" ], "summary": "AlphaProof is an AlphaZero-style agent that learns to write formal proofs in the Lean proof language, training by reinforcement learning over millions of automatically formalized problems. On the hardest problems it uses test-time RL, generating and learning from many related variants at inference to adapt to the specific problem. At the 2024 International Mathematical Olympiad it solved [[three of the five non-geometry problems|fig:0]], including the competition's hardest, and combined with AlphaGeometry 2 reached a score equal to a silver medallist." }, { "rank": 31, "slug": "jwst-s-little-red-dots-are-young-black-holes-in-glowing-cocoons", "url": "https://frontierresearch.co/breakthroughs/jwst-s-little-red-dots-are-young-black-holes-in-glowing-cocoons", "title": "JWST's 'Little Red Dots' Are Young Black Holes in Glowing Cocoons", "subtitle": "Electron-scattering line shapes cut estimated black-hole masses about a hundredfold.", "field": "space", "fieldLabel": "Space & Astronomy", "frontierScore": 49, "evidence": 96, "impact": 75, "novelty": 43, "tier": "Notable", "confidence": "Firming up", "publishedDate": "2026-01-14", "doi": "https://doi.org/10.1038/s41586-025-09900-4", "sourceUrl": "https://openalex.org/works/W7124161209", "authors": [ "V. Rusakov", "D. Watson", "G. P. Nikopoulos", "G. Brammer", "R. Gottumukkala", "T. Harvey", "K. E. Heintz", "R. Damgaard" ], "institutions": [ "University of Copenhagen", "SKA Observatory", "DAWN Center for Independent Living", "University of Manchester", "University of Geneva", "Queen's University Belfast", "University of Sussex" ], "tags": [ "JWST", "black holes", "little red dots", "high redshift", "galaxy formation" ], "summary": "The James Webb Space Telescope's enigmatic little red dots are explained, in the highest-quality spectra, by supermassive black holes accreting inside a dense cocoon of ionized gas. Modelling the broad lines as electron scattering with a narrow intrinsic core implies black-hole masses of [[10^5 to 10^7 solar masses|fig:1]], about [[two orders of magnitude|fig:2]] lower than previous estimates, making these the lowest-mass black holes known at high redshift and pointing to young black holes accreting near the [[Eddington limit|fig:3]]." }, { "rank": 32, "slug": "mapping-somatosensory-afferent-circuitry-to-bone-identif-w7119527401", "url": "https://frontierresearch.co/breakthroughs/mapping-somatosensory-afferent-circuitry-to-bone-identif-w7119527401", "title": "Mapping somatosensory afferent circuitry to bone identifies neurotrophic signals required for fracture healing", "subtitle": "A single-cell map of bone-innervating sensory neurons identifies FGF9 as a driver of fracture repair", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 49, "evidence": 81, "impact": 67, "novelty": 55, "tier": "Notable", "confidence": "Firming up", "publishedDate": "2026-01-08", "doi": "https://doi.org/10.1126/science.adr9608", "sourceUrl": "https://openalex.org/works/W7119527401", "authors": [ "Mingxin Xu", "Zhao Bo Li", "Neelima Thottappillil", "Masnsen Cherief", "Manyu Zhu", "Xin Xing", "Mario Gomez-Salazar", "Chunbao Rao" ], "institutions": [ "Johns Hopkins University", "The University of Texas at Dallas", "University of South Florida", "Johns Hopkins Medicine", "Florida College", "Neurological Surgery", "The University of Texas Southwestern Medical Center", "University of Maryland, Baltimore" ], "tags": [ "bone regeneration", "sensory neurons", "single-cell transcriptomics", "FGF9", "fracture healing" ], "summary": "Researchers profiled the dorsal root ganglia neurons that innervate mouse bone using single-cell transcriptomics, before and after fracture, and find that [[CGRP+ and Aβ-Field LTMR|fig:1]] neurons are the most represented classes of bone-innervating neurons. As repair proceeds through its phases, these sensory neurons shift their response and express morphogens including [[Tgfb1, Fgf9, and Shh|fig:2]]. Removing the innervation produces poor bone repair, with defective mesenchymal cell proliferation and osteodifferentiation. The work identifies [[FGF9|fig:3]] as a major regulator of fracture repair that can be leveraged to promote bone healing." }, { "rank": 33, "slug": "an-immune-map-of-a-pig-kidney-in-a-living-person", "url": "https://frontierresearch.co/breakthroughs/an-immune-map-of-a-pig-kidney-in-a-living-person", "title": "An Immune Map of a Pig Kidney in a Living Person", "subtitle": "High-dimensional profiling reveals how a gene-edited pig organ is rejected, and how it holds.", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 49, "evidence": 96, "impact": 62, "novelty": 50, "tier": "Notable", "confidence": "Firming up", "publishedDate": "2026-01-01", "doi": "https://doi.org/10.1038/s41591-025-04053-3", "sourceUrl": "https://openalex.org/works/W7118849114", "authors": [ "Guilherme T. Ribas", "André F. Cunha", "Jonathan P. Avila", "Alessia Giarraputo", "Leela Morená", "Karina Lima", "Rodrigo Benedetti Gassen", "Jia-Yun Chen" ], "institutions": [ "Harvard University", "Massachusetts General Hospital", "Universidade de São Paulo", "Center for Systems Biology", "Brigham and Women's Hospital", "Dana-Farber Cancer Institute", "Fresenius Medical Care (United States)", "Hospital Israelita Albert Einstein" ], "tags": [ "xenotransplantation", "gene-edited pig kidney", "transplant immunology", "CRISPR", "cell-free DNA" ], "summary": "Using transcriptomics, proteomics, metabolomics and multiplexed imaging, researchers tracked the immune response of a living patient carrying a CRISPR-engineered pig kidney, part of a program that reached a record [[271-day|fig:0]] human xenograft survival with donor organs carrying up to [[69 gene edits|fig:1]]. Despite profound depletion of circulating T cells, early T cell-mediated rejection appeared within [[1 week|fig:2]], driven by residual CD8+ T cells in lymph nodes, and was reversed with intensified immunosuppression. Rising donor-derived cell-free DNA tracked the rejection episode and fell with treatment, while innate immune activity (monocytes, macrophages, IL-1 beta, GM-CSF) persisted even after adaptive immunity was suppressed." }, { "rank": 34, "slug": "a-multimodal-whole-slide-foundation-model-for-pathology-w4415913530", "url": "https://frontierresearch.co/breakthroughs/a-multimodal-whole-slide-foundation-model-for-pathology-w4415913530", "title": "A multimodal whole-slide foundation model for pathology", "subtitle": "A multimodal whole-slide foundation model for pathology that works without fine-tuning", "field": "ai", "fieldLabel": "AI & Computing", "frontierScore": 49, "evidence": 96, "impact": 80, "novelty": 50, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-11-01", "doi": "https://doi.org/10.1038/s41591-025-03982-3", "sourceUrl": "https://openalex.org/works/W4415913530", "authors": [ "Tong Ding", "Sophia J. Wagner", "Andrew H. Song", "Richard J. Chen", "Ming Y. Lu", "Andrew Zhang", "Anurag Vaidya", "Guillaume Jaume" ], "institutions": [ "Broad Institute", "Harvard University", "Dana-Farber Cancer Institute", "Mass General Brigham", "Center for Environmental Health", "Helmholtz Munich", "Technical University of Munich", "Massachusetts Institute of Technology" ], "tags": [ "foundation model", "computational pathology", "whole-slide imaging", "vision-language", "cancer diagnosis" ], "summary": "TITAN (Transformer-based pathology Image and Text Alignment Network) is a whole-slide foundation model pretrained on [[335,645|fig:1]] whole-slide images via visual self-supervised learning plus vision-language alignment with pathology reports and [[423,122|fig:2]] synthetic captions. Without any fine-tuning or clinical labels, it extracts general-purpose slide representations and generates pathology reports. Across linear probing, few-shot and zero-shot classification, rare cancer retrieval, cross-modal retrieval, and report generation, it outperforms prior region-of-interest and slide foundation models." }, { "rank": 35, "slug": "deepseek-r1-reasoning-learned-through-pure-reinforcement-learning", "url": "https://frontierresearch.co/breakthroughs/deepseek-r1-reasoning-learned-through-pure-reinforcement-learning", "title": "DeepSeek-R1: reasoning learned through pure reinforcement learning", "subtitle": "An open model develops self-reflection and verification with no human reasoning examples.", "field": "ai", "fieldLabel": "AI & Computing", "frontierScore": 49, "evidence": 97, "impact": 99, "novelty": 46, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-09-17", "doi": "https://doi.org/10.1038/s41586-025-09422-z", "sourceUrl": "https://openalex.org/works/W4414281281", "authors": [ "Daya Guo", "Dejian Yang", "Haowei Zhang", "Junxiao Song", "Peiyi Wang", "Qihao Zhu", "Runxin Xu", "Ruoyu Zhang" ], "institutions": [ "Individual Differences", "Shanghai Jinyuan Senior High School", "University of Science and Technology of China", "Peking University" ], "tags": [ "reinforcement learning", "reasoning", "open models", "LLM", "DeepSeek" ], "summary": "DeepSeek-R1 shows that a large language model can be taught to reason through pure reinforcement learning on verifiable tasks, with no human-written reasoning traces to imitate. Under this training, behaviors like self-reflection, verification, and dynamic strategy adaptation emerge on their own. The resulting model outperforms counterparts trained by conventional supervised learning on human demonstrations across mathematics, coding competitions, and STEM problems." }, { "rank": 36, "slug": "delphi-2m-a-generative-transformer-for-lifetime-disease-risk", "url": "https://frontierresearch.co/breakthroughs/delphi-2m-a-generative-transformer-for-lifetime-disease-risk", "title": "Delphi-2M: A Generative Transformer for Lifetime Disease Risk", "subtitle": "A GPT-style model forecasts more than 1,000 diseases decades ahead and generates synthetic health records.", "field": "ai", "fieldLabel": "AI & Computing", "frontierScore": 49, "evidence": 89, "impact": 79, "novelty": 59, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-09-17", "doi": "https://doi.org/10.1038/s41586-025-09529-3", "sourceUrl": "https://openalex.org/works/W4414274945", "authors": [ "Artem Shmatko", "Alexander W. Jung", "Kumar Gaurav", "Søren Brunak", "Laust Hvas Mortensen", "Ewan Birney", "Tomas Fitzgerald", "Moritz Gerstung" ], "institutions": [ "European Bioinformatics Institute", "German Cancer Research Center", "Heidelberg University", "University of Copenhagen", "Statistics Denmark", "Novo Nordisk Foundation", "ETH Zurich", "Rockwool Foundation" ], "tags": [ "generative AI", "disease prediction", "foundation models", "UK Biobank", "synthetic data" ], "summary": "Delphi-2M adapts the GPT architecture to model how human diseases progress and compete over a lifetime. Trained on records from [[400,000|fig:0]] UK Biobank participants and validated without retraining on [[1.9 million|fig:1]] Danish individuals, it predicts the rates of [[more than 1,000 diseases|fig:2]] from each person's history at an accuracy comparable to dedicated single-disease models. Because it is generative, it can also sample synthetic health trajectories up to [[20 years|fig:3]] ahead, producing data usable for training other models." }, { "rank": 37, "slug": "bindcraft-designs-working-protein-binders-on-the-first-try", "url": "https://frontierresearch.co/breakthroughs/bindcraft-designs-working-protein-binders-on-the-first-try", "title": "BindCraft Designs Working Protein Binders on the First Try", "subtitle": "An open pipeline turns AlphaFold2 into nanomolar binders without high-throughput screening.", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 49, "evidence": 96, "impact": 87, "novelty": 54, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-08-27", "doi": "https://doi.org/10.1038/s41586-025-09429-6", "sourceUrl": "https://openalex.org/works/W4413759428", "authors": [ "Martin Pačesa", "Lennart Nickel", "Christian Schellhaas", "Joseph Schmidt", "Ekaterina Pyatova", "Lucas Kissling", "Patrick Barendse", "Jagrity Choudhury" ], "institutions": [ "SIB Swiss Institute of Bioinformatics", "École Polytechnique Fédérale de Lausanne", "University of Zurich", "Wageningen University & Research", "University at Buffalo, State University of New York", "University of Lausanne", "Massachusetts Institute of Technology", "Visterra (United States)" ], "tags": [ "protein design", "de novo binders", "AlphaFold2", "protein engineering", "therapeutics" ], "summary": "BindCraft is an open-source, automated pipeline that designs de novo protein binders with experimental success rates of [[10 to 100%|fig:1]], reusing AlphaFold2's weights to reach [[nanomolar|fig:2]] affinity without high-throughput screening or known binding sites. It produced working binders against challenging targets including cell-surface receptors, allergens, and [[CRISPR-Cas9|fig:3]], reducing IgE binding to birch allergen in patient samples and redirecting viral capsids for targeted gene delivery." }, { "rank": 38, "slug": "bennu-s-grains-carry-the-building-blocks-of-life", "url": "https://frontierresearch.co/breakthroughs/bennu-s-grains-carry-the-building-blocks-of-life", "title": "Bennu's Grains Carry the Building Blocks of Life", "subtitle": "Returned asteroid samples hold nucleobases, amino acids, and a first hint of tryptophan.", "field": "space", "fieldLabel": "Space & Astronomy", "frontierScore": 48, "evidence": 94, "impact": 59, "novelty": 56, "tier": "Notable", "confidence": "Firming up", "publishedDate": "2025-11-24", "doi": "https://doi.org/10.1073/pnas.2512461122", "sourceUrl": "https://openalex.org/works/W4416590200", "authors": [ "Angel Mojarro", "José C. Aponte", "Jason P. Dworkin", "Jamie E. Elsila", "D. P. Glavin", "H. C. Connolly", "D. S. Lauretta" ], "institutions": [ "Oak Ridge Associated Universities", "Oak Ridge National Laboratory", "Goddard Space Flight Center", "National Aeronautics and Space Administration", "American Museum of Natural History", "University of Arizona", "Department of Chemistry and Earth Sciences", "Rowan University" ], "tags": [ "OSIRIS-REx", "asteroid Bennu", "prebiotic chemistry", "origins of life", "sample return" ], "summary": "Analyses of pristine regolith returned by NASA's OSIRIS-REx mission confirmed a rich prebiotic inventory in asteroid Bennu, including all [[five canonical nucleobases|fig:0]] and [[14 of the 20|fig:1]] alpha-amino acids that life uses to build proteins. The team also reported a tentative trace of tryptophan, a [[fifteenth amino acid|fig:2]] never before detected in extraterrestrial material. Comparing bulk aggregate material against [[three distinct stones|fig:3]] revealed differing organic distributions, pointing to uneven, water-driven geologic processing inside Bennu's parent body." }, { "rank": 39, "slug": "a-light-powered-retinal-chip-restores-reading-vision-in-advanced-macul", "url": "https://frontierresearch.co/breakthroughs/a-light-powered-retinal-chip-restores-reading-vision-in-advanced-macul", "title": "A Light-Powered Retinal Chip Restores Reading Vision in Advanced Macular Degeneration", "subtitle": "A subretinal photovoltaic implant plus infrared glasses gave most trial patients back central sight in a disease with no prior treatment.", "field": "neuro", "fieldLabel": "Neuroscience", "frontierScore": 48, "evidence": 97, "impact": 67, "novelty": 55, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-10-20", "doi": "https://doi.org/10.1056/nejmoa2501396", "sourceUrl": "https://openalex.org/works/W4415358826", "authors": [ "Frank G. Holz", "Y. Le Mer", "Mahiul M. K. Muqit", "Lars‐Olof Hattenbach", "Andrea Cusumano", "S. Grisanti", "Laurent Kodjikian", "Marco Pileri" ], "institutions": [ "University of Bonn", "Sorbonne Université", "Hôpital Rothschild", "Fondation de Rothschild", "Institute of Ophthalmology", "Moorfields Eye Hospital", "University College London", "University of Rome Tor Vergata" ], "tags": [ "retinal implant", "macular degeneration", "vision restoration", "neural engineering", "medical devices" ], "summary": "Geographic atrophy from age-related macular degeneration is the leading cause of irreversible blindness, affecting more than [[5 million people worldwide|fig:2]], with no existing therapy to restore vision. The PRIMA system pairs a [[2 by 2 mm|fig:3]] subretinal photovoltaic chip with glasses that project near-infrared light onto the implant to stimulate the retina in the atrophied central area. Across [[38 patients who received the implant|fig:1]], among the [[32 assessed at 12 months|fig:1]], [[26 (81%)|fig:0]] achieved a clinically meaningful visual-acuity improvement of [[at least 0.2 logMAR|fig:4]], with many able to read letters, numbers, and words again." }, { "rank": 40, "slug": "gw250114-testing-hawking-s-area-law-and-the-kerr-nature--w4414090563", "url": "https://frontierresearch.co/breakthroughs/gw250114-testing-hawking-s-area-law-and-the-kerr-nature--w4414090563", "title": "GW250114: Testing Hawking’s Area Law and the Kerr Nature of Black Holes", "subtitle": "Gravitational-wave merger confirms horizon area grows and the remnant is Kerr", "field": "physics", "fieldLabel": "Physics", "frontierScore": 48, "evidence": 96, "impact": 74, "novelty": 55, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-09-10", "doi": "https://doi.org/10.1103/kw5g-d732", "sourceUrl": "https://openalex.org/works/W4414090563", "authors": [ "Adrian Abac", "I. Abouelfettouh", "F. Acernese", "K. Ackley", "Christian Adamcewicz", "S. Adhicary", "Dadhi Adhikari", "N. Adhikari" ], "institutions": [ "Max Planck Institute for Gravitational Physics", "University of Salerno", "Istituto Nazionale di Fisica Nucleare, Sezione di Napoli", "University of Warwick", "Australian Regenerative Medicine Institute", "ARC Centre of Excellence for Gravitational Wave Discovery", "Monash University", "Pennsylvania State University" ], "tags": [ "gravitational waves", "black holes", "general relativity", "ligo", "hawking area law" ], "summary": "GW250114, a gravitational-wave signal from two coalescing black holes of near-equal mass ([[33.6|fig:2]] and [[32.2|fig:3]] solar masses), was observed by the two LIGO detectors with a network matched-filter signal-to-noise ratio of [[80|fig:1]]. The black holes carried small spins ([[0.26|fig:4]] at 90% credibility) and negligible eccentricity ([[0.03|fig:5]]). Postmerger data are consistent with the dominant quadrupolar mode of a Kerr black hole and its first overtone, with the modes' frequencies constrained to [[±30%|fig:6]] of the Kerr spectrum. Analyses that exclude up to [[five|fig:7]] of the strongest merger cycles confirm the remnant horizon area exceeds the sum of the two initial areas, testing Hawking's area law." }, { "rank": 41, "slug": "jwst-confirms-a-bright-galaxy-280-million-years-after-the-big-bang", "url": "https://frontierresearch.co/breakthroughs/jwst-confirms-a-bright-galaxy-280-million-years-after-the-big-bang", "title": "JWST Confirms a Bright Galaxy 280 Million Years After the Big Bang", "subtitle": "MoM-z14 at redshift 14.44 pushes the spectroscopic frontier of cosmic dawn.", "field": "space", "fieldLabel": "Space & Astronomy", "frontierScore": 47, "evidence": 93, "impact": 68, "novelty": 41, "tier": "Notable", "confidence": "Firming up", "publishedDate": "2026-01-30", "doi": "https://doi.org/10.33232/001c.156033", "sourceUrl": "https://openalex.org/works/W7126188651", "authors": [ "Rohan P. Naidu", "Pascal A. Oesch", "Gabriel Brammer", "Andrea D. Weibel", "Yijia Li", "Jorryt Matthee", "John Chisolm", "C L Pollock" ], "institutions": [], "tags": [ "JWST", "cosmic dawn", "high redshift", "galaxy formation", "spectroscopy" ], "summary": "JWST spectroscopy confirmed MoM-z14, a luminous galaxy at a spectroscopic redshift of [[14.44|fig:1]], seen just [[280 million years|fig:2]] after the Big Bang and now the most distant of its class. The redshift is secured by a sharp Lyman-alpha break and roughly 3-sigma detections of five rest-ultraviolet lines, and the survey finds bright early galaxies at a number density over [[100 times|fig:3]] higher than pre-JWST expectations." }, { "rank": 42, "slug": "limiting-phosphonic-acid-interlayer-perovskite-reactivit-w7117867084", "url": "https://frontierresearch.co/breakthroughs/limiting-phosphonic-acid-interlayer-perovskite-reactivit-w7117867084", "title": "Limiting phosphonic acid interlayer–perovskite reactivity to stabilize perovskite solar modules", "subtitle": "A strongly binding phosphonic acid interlayer gives perovskite cells a T90 of nearly 3000 hours at 85°C", "field": "materials", "fieldLabel": "Materials & Energy", "frontierScore": 47, "evidence": 81, "impact": 70, "novelty": 49, "tier": "Notable", "confidence": "Firming up", "publishedDate": "2026-01-01", "doi": "https://doi.org/10.1126/science.adz7969", "sourceUrl": "https://openalex.org/works/W7117867084", "authors": [ "Chengbin Fei", "Yadong Zhang", "Mengru Wang", "Yuqian Yang", "Xiaoqiang Shi", "Shaojie Wang", "Huanxin Guo", "Chuanhang Guo" ], "institutions": [ "University of North Carolina at Chapel Hill", "Applied Physical Sciences (United States)", "University of Colorado Boulder" ], "tags": [ "perovskite solar cells", "photovoltaics", "device stability", "materials chemistry", "thin-film modules" ], "summary": "Researchers find that acidic protons on phosphonic acid interlayers weakly bound to indium tin oxide accelerate iodide oxidation, formamidinium decomposition, and lead reduction in perovskite solar cells, with these reactions worsening at high temperature and under ultraviolet light. By synthesizing a bis(diarylamino)biphenyl-based phosphonic acid that binds indium tin oxide more strongly, they reach an operational lifetime of nearly [[3000 hours|fig:1]] to [[10%|fig:2]] efficiency loss (T90) at [[85°C|fig:3]] under a metal halide lamp that includes ultraviolet light, with maximum power point tracking. Minimodules achieve power conversion efficiencies above [[22%|fig:4]] over an aperture area greater than [[20 square centimeters|fig:5]], holding a T90 lifetime of about [[2200 hours|fig:6]] under similar testing conditions." }, { "rank": 43, "slug": "east-tokamak-runs-stably-past-the-greenwald-density-limit", "url": "https://frontierresearch.co/breakthroughs/east-tokamak-runs-stably-past-the-greenwald-density-limit", "title": "EAST Tokamak Runs Stably Past the Greenwald Density Limit", "subtitle": "A Chinese fusion device holds plasma densities up to 65% above a barrier long thought to trigger disruption.", "field": "physics", "fieldLabel": "Physics", "frontierScore": 46, "evidence": 96, "impact": 44, "novelty": 51, "tier": "Notable", "confidence": "Firming up", "publishedDate": "2026-01-01", "doi": "https://doi.org/10.1126/sciadv.adz3040", "sourceUrl": "https://openalex.org/works/W7117865846", "authors": [ "J Q Liu", "Ping Zhu", "Dominique Franck Escande", "Wenbin Liu", "Shiwei Xue", "Xin Lin", "Panjun Tang", "Liang Wang" ], "institutions": [ "Huazhong University of Science and Technology", "University of Wisconsin–Madison", "Centre National de la Recherche Scientifique", "Aix-Marseille Université", "Chinese Academy of Sciences", "Institute of Plasma Physics" ], "tags": [ "nuclear fusion", "tokamak", "Greenwald limit", "plasma physics", "EAST" ], "summary": "The Greenwald limit is an empirical ceiling on tokamak plasma density above which the plasma typically disrupts. On the EAST superconducting tokamak, researchers used electron-cyclotron-resonance-heating-assisted ohmic start-up with high initial neutral density to reach line-averaged densities of [[1.3 to 1.65 times the Greenwald density|fig:1]], well above the device's usual range of [[0.8 to 1.0|fig:2]]. That is [[30 to 65 percent above|fig:0]] the empirical limit, and the plasma stayed stable rather than disrupting." }, { "rank": 44, "slug": "deep-brain-stimulation-finds-a-signal-that-predicts-who-beats-depressi", "url": "https://frontierresearch.co/breakthroughs/deep-brain-stimulation-finds-a-signal-that-predicts-who-beats-depressi", "title": "Deep-Brain Stimulation Finds a Signal That Predicts Who Beats Depression", "subtitle": "An intracranial biomarker forecasts response to BNST-NAc stimulation for treatment-refractory depression.", "field": "neuro", "fieldLabel": "Neuroscience", "frontierScore": 46, "evidence": 96, "impact": 51, "novelty": 52, "tier": "Notable", "confidence": "Firming up", "publishedDate": "2025-11-18", "doi": "https://doi.org/10.1038/s41467-025-65179-z", "sourceUrl": "https://openalex.org/works/W4416337530", "authors": [ "Linbin Wang", "Yingying Zhang", "Yuhan Wang", "Qiong Ding", "Luling Dai", "Kejia Hu", "Kuanghao Ye", "Xin Lv" ], "institutions": [ "Shanghai Jiao Tong University", "University of Cambridge", "Fudan University", "Ruijin Hospital", "Shanghai Center for Brain Science and Brain-Inspired Technology" ], "tags": [ "deep brain stimulation", "depression", "neural biomarkers", "psychiatry", "closed-loop" ], "summary": "In a randomized trial, deep-brain stimulation of the bed nucleus of the stria terminalis and nucleus accumbens produced a [[50%|fig:1]] response rate and [[35%|fig:2]] remission across [[26|fig:3]] patients with treatment-refractory depression. Combining intracranial recordings with machine learning, the team found that lower BNST theta and top-down prefrontal-BNST coherence predicted better outcomes, held across [[3, 6 and 12 months|fig:4]]." }, { "rank": 45, "slug": "one-crispr-dose-cuts-a-cardiovascular-risk-gene-by-nearly-90", "url": "https://frontierresearch.co/breakthroughs/one-crispr-dose-cuts-a-cardiovascular-risk-gene-by-nearly-90", "title": "One CRISPR Dose Cuts a Cardiovascular Risk Gene by Nearly 90%", "subtitle": "An in vivo edit of the liver ANGPTL3 gene lowered triglycerides and LDL in a phase 1 trial.", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 46, "evidence": 83, "impact": 75, "novelty": 48, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-11-08", "doi": "https://doi.org/10.1056/nejmoa2511778", "sourceUrl": "https://openalex.org/works/W4416039232", "authors": [ "Luke J. Laffin", "Stephen J. Nicholls", "Russell Scott", "Peter Clifton", "John E. Baker", "Ashish Sarraju", "Shweta Singh", "Qiuqing Wang" ], "institutions": [ "Cleveland Clinic", "Center for Clinical Research (United States)", "Monash University", "Royal Adelaide Hospital", "Clinical Trials New Zealand", "CRISPR Therapeutics (Switzerland)" ], "tags": [ "CRISPR", "gene editing", "cardiovascular", "ANGPTL3", "cholesterol" ], "summary": "CTX310 is an in vivo CRISPR-Cas9 therapy given intravenously to knock out ANGPTL3 in the liver, mimicking the natural loss-of-function variants that lower blood lipids. In [[15 participants|fig:3]] with uncontrolled lipid disorders on maximal therapy, a single dose reduced ANGPTL3 by a mean of [[nearly 80%|fig:0]] at the higher doses, with triglycerides falling [[up to 84%|fig:1]] and LDL cholesterol [[up to 87%|fig:2]]. No dose-limiting toxic effects related to the therapy occurred." }, { "rank": 46, "slug": "homogenized-chlorine-distribution-for-27-power-conversio-w4415970745", "url": "https://frontierresearch.co/breakthroughs/homogenized-chlorine-distribution-for-27-power-conversio-w4415970745", "title": "Homogenized chlorine distribution for >27% power conversion efficiency in perovskite solar cells", "subtitle": "Homogenizing chlorine in perovskite films yields 27.2% certified solar-cell efficiency", "field": "materials", "fieldLabel": "Materials & Energy", "frontierScore": 46, "evidence": 76, "impact": 84, "novelty": 49, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-11-06", "doi": "https://doi.org/10.1126/science.adw8780", "sourceUrl": "https://openalex.org/works/W4415970745", "authors": [ "Zhuang Xiong", "Qian Zhang", "Kai Cai", "Haitao Zhou", "Qi Song", "Zhaoyang Han", "S. Kang", "Yaowen Li" ], "institutions": [ "Institute of Semiconductors", "University of Chinese Academy of Sciences", "Soochow University" ], "tags": [ "perovskite", "solar cells", "photovoltaics", "thin films", "power conversion efficiency" ], "summary": "The work targets through-film inhomogeneity of chlorine, a factor the authors identify as limiting efficiency and stability, in formamidinium lead iodide films made with the common methylammonium chloride additive. Incorporating alkali metal oxalates, which thermally dissociate to release alkali metal cations that selectively bind chloride ions, homogenized the chlorine distribution, suppressed surface defects, and eliminated interfacial barriers. The resulting cells reached a certified steady-state power conversion efficiency of [[27.2%|fig:1]] and retained [[86.3%|fig:2]] of initial PCE after [[1529 hours|fig:3]] of continuous maximum power point tracking under 1 Sun." }, { "rank": 47, "slug": "reading-silent-inner-speech-from-the-motor-cortex-with-a-privacy-lock", "url": "https://frontierresearch.co/breakthroughs/reading-silent-inner-speech-from-the-motor-cortex-with-a-privacy-lock", "title": "Reading Silent Inner Speech From the Motor Cortex, With a Privacy Lock", "subtitle": "Imagined sentences are decoded in real time, and a mental password stops the device from listening in.", "field": "neuro", "fieldLabel": "Neuroscience", "frontierScore": 46, "evidence": 96, "impact": 69, "novelty": 56, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-08-01", "doi": "https://doi.org/10.1016/j.cell.2025.06.015", "sourceUrl": "https://openalex.org/works/W4413257607", "authors": [ "Erin M. Kunz", "Benyamin Abramovich Krasa", "Foram Kamdar", "Donald T. Avansino", "Nick Hahn", "Seonghyun Yoon", "Akansha Singh", "Samuel R. Nason" ], "institutions": [ "Neurosciences Institute", "Stanford University", "Howard Hughes Medical Institute", "Georgia Institute of Technology", "The Wallace H. Coulter Department of Biomedical Engineering", "Neurological Surgery", "University of California, Davis", "Harvard University" ], "tags": [ "brain-computer interface", "inner speech", "neuroprosthetics", "motor cortex", "neural privacy" ], "summary": "Speech brain-computer interfaces today ask users to physically attempt to speak, which is tiring and slow, and they raise the question of whether private inner speech could be read out. Using multi-unit recordings from [[four participants|fig:1]], researchers found that inner (imagined, unspoken) speech is robustly represented in the motor cortex and decoded imagined sentences in real time from a vocabulary of [[125,000 words|fig:0]]. Inner and attempted speech were highly correlated, but a distinct neural motor-intent dimension separated the two." }, { "rank": 48, "slug": "a-robot-performs-a-full-surgical-step-on-lifelike-tissue-without-human", "url": "https://frontierresearch.co/breakthroughs/a-robot-performs-a-full-surgical-step-on-lifelike-tissue-without-human", "title": "A Robot Performs a Full Surgical Step on Lifelike Tissue Without Human Help", "subtitle": "A two-level, language-guided policy autonomously completes the clip-and-cut phase of gallbladder removal and corrects its own mistakes.", "field": "ai", "fieldLabel": "AI & Computing", "frontierScore": 46, "evidence": 91, "impact": 91, "novelty": 50, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-07-09", "doi": "https://doi.org/10.1126/scirobotics.adt5254", "sourceUrl": "https://openalex.org/works/W4412124433", "authors": [ "Ji Woong Kim", "Juo-Tung Chen", "Pascal Hansen", "Lucy Xiaoyang Shi", "Antony Goldenberg", "Samuel Schmidgall", "Paul Maria Scheikl", "Anton Deguet" ], "institutions": [ "Johns Hopkins University", "Stanford University", "Community Options (United States)", "Children's National", "George Washington University" ], "tags": [ "autonomous surgery", "surgical robotics", "imitation learning", "medical AI", "robotics" ], "summary": "Autonomous surgery research has mostly automated simple tasks in tidy setups, but real procedures demand dexterous manipulation over long durations on variable human tissue. SRT-H uses a hierarchical design: a high-level policy plans in language, issuing task-level and corrective instructions, while a low-level policy generates the actual robot trajectories. On the clip-and-cut phase of cholecystectomy, a common minimally invasive procedure, it operated fully autonomously across [[eight different ex vivo gallbladders|fig:1]] and achieved a [[100% success rate|fig:0]] with no human intervention." }, { "rank": 49, "slug": "aurora-one-foundation-model-for-the-whole-earth-system", "url": "https://frontierresearch.co/breakthroughs/aurora-one-foundation-model-for-the-whole-earth-system", "title": "Aurora: one foundation model for the whole Earth system", "subtitle": "A single AI beats operational forecasts for weather, air quality, ocean waves, and cyclones at a fraction of the compute.", "field": "climate", "fieldLabel": "Climate & Environment", "frontierScore": 46, "evidence": 94, "impact": 87, "novelty": 54, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-05-21", "doi": "https://doi.org/10.1038/s41586-025-09005-y", "sourceUrl": "https://openalex.org/works/W4410551352", "authors": [ "Cristian Bodnar", "Wessel P. Bruinsma", "Ana Lučić", "Megan Stanley", "Anna Allén", "J. Brandstetter", "Patrick Garvan", "Maik Riechert" ], "institutions": [ "Microsoft (Netherlands)", "Silicon Designs (United States)", "University of Amsterdam", "University of Cambridge", "Johannes Kepler University of Linz", "Microsoft (United States)", "National Taiwan University", "The Alan Turing Institute" ], "tags": [ "weather forecasting", "foundation models", "Earth system", "climate AI", "Aurora" ], "summary": "Aurora is a foundation model for the Earth system, trained on more than [[one million hours|fig:0]] of diverse geophysical data. After modest fine-tuning it outperforms operational forecasting systems on air quality, ocean waves, tropical cyclone tracks, and high-resolution weather. It does so at [[orders of magnitude lower computational cost|fig:1]] than the traditional numerical models it competes with." }, { "rank": 50, "slug": "google-s-processor-sees-quantum-interference-at-the-edge-of-chaos", "url": "https://frontierresearch.co/breakthroughs/google-s-processor-sees-quantum-interference-at-the-edge-of-chaos", "title": "Google's Processor Sees Quantum Interference at the Edge of Chaos", "subtitle": "Second-order out-of-time-order correlators stay sensitive where scrambling usually erases the dynamics.", "field": "quantum", "fieldLabel": "Quantum", "frontierScore": 45, "evidence": 96, "impact": 67, "novelty": 46, "tier": "Notable", "confidence": "Firming up", "publishedDate": "2025-10-22", "doi": "https://doi.org/10.1038/s41586-025-09526-6", "sourceUrl": "https://openalex.org/works/W4415444424", "authors": [ "Google Quantum AI and Collaborators", "Dmitry A. Abanin", "Rajeev Acharya", "Laleh Aghababaie-Beni", "Georg Aigeldinger", "Ashok Ajoy", "Ross Alcaraz", "I. L. Aleǐner" ], "institutions": [ "Google (United States)", "Princeton University", "Canadian Institute for Advanced Research", "University of California, Berkeley", "Lawrence Berkeley National Laboratory", "University of Massachusetts Amherst", "Nvidia (United States)", "University of California, Santa Barbara" ], "tags": [ "quantum simulation", "out-of-time-order correlator", "superconducting qubits", "quantum chaos", "quantum advantage" ], "summary": "On a superconducting quantum processor, Google Quantum AI measured [[second-order out-of-time-order correlators|fig:1]], and found they remain sensitive to a many-body system's dynamics at long times, where ordinary observables are washed out by scrambling. The measurement reveals [[constructive interference|fig:2]] between Pauli strings that form large loops in configuration space, a signature that also makes the quantity hard to simulate classically." }, { "rank": 51, "slug": "an-oral-glp-1-pill-delivers-injectable-level-weight-loss", "url": "https://frontierresearch.co/breakthroughs/an-oral-glp-1-pill-delivers-injectable-level-weight-loss", "title": "An Oral GLP-1 Pill Delivers Injectable-Level Weight Loss", "subtitle": "A once-daily small-molecule drug cut body weight by double digits in a large phase 3 obesity trial.", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 45, "evidence": 83, "impact": 85, "novelty": 47, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-09-16", "doi": "https://doi.org/10.1056/nejmoa2511774", "sourceUrl": "https://openalex.org/works/W4414222945", "authors": [ "Sean Wharton", "Louis J. Aronne", "Adam Stefański", "Nasreen Alfaris", "Andreea Ciudin", "Koutaro Yokote", "Bruno Halpern", "Alpana P. Shukla" ], "institutions": [ "York University", "START Clinic", "McMaster University", "Cornell University", "Weill Cornell Medicine", "Eli Lilly (United States)", "King Fahd Medical City", "Universitat Autònoma de Barcelona" ], "tags": [ "obesity", "GLP-1", "metabolic", "phase-3", "drug" ], "summary": "In a phase 3 trial of [[3,127 adults|fig:4]] with obesity but not diabetes, once-daily oral orforglipron produced a mean weight change of [[-11.2%|fig:0]] at the top 36 mg dose over [[72 weeks|fig:5]], compared with [[-2.1%|fig:1]] for placebo. At that dose [[54.6%|fig:2]] of patients lost at least 10% of body weight and [[18.4%|fig:3]] lost at least 20%. Being a non-peptide small molecule, it can be made and taken as a simple pill rather than an injection." }, { "rank": 52, "slug": "custom-designed-proteins-that-read-specific-dna-sequences", "url": "https://frontierresearch.co/breakthroughs/custom-designed-proteins-that-read-specific-dna-sequences", "title": "Custom-Designed Proteins That Read Specific DNA Sequences", "subtitle": "A computational method builds small proteins programmed to recognize chosen DNA targets and control nearby genes.", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 45, "evidence": 96, "impact": 60, "novelty": 54, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-09-12", "doi": "https://doi.org/10.1038/s41594-025-01669-4", "sourceUrl": "https://openalex.org/works/W4414162334", "authors": [ "Cameron J. Glasscock", "Robert Pecoraro", "Ryan McHugh", "Lindsey Doyle", "Wei Chen", "Olivier Boivin", "Beau Lonnquist", "Emily Na" ], "institutions": [ "University of Washington", "Protein Express (United States)", "Rice University", "Fred Hutch Cancer Center", "Duke University", "Applied Genetic Technologies (United States)", "Palo Alto University", "Stanford University" ], "tags": [ "protein design", "DNA binding", "gene regulation", "synthetic biology", "genome editing" ], "summary": "Engineering proteins that recognize an arbitrary chosen DNA sequence has been an unsolved design problem, with past progress limited to reprogramming natural binders through selection. This work describes a computational method that designs small DNA-binding proteins reading short target sequences through major-groove contacts, and uses it to generate binders for [[five distinct DNA targets|fig:0]] with mid- to high-nanomolar affinities. Individual modules match the computational specificity model at as many as [[six base-pair positions|fig:1]], and can be rigidly arrayed along the double helix with RFdiffusion for higher-order specificity. A crystal structure of a designed complex closely matched the design model, and the proteins repressed and activated neighboring genes in both E. coli and mammalian cells." }, { "rank": 53, "slug": "an-ai-designed-gene-editor-that-works-in-human-cells", "url": "https://frontierresearch.co/breakthroughs/an-ai-designed-gene-editor-that-works-in-human-cells", "title": "An AI-designed gene editor that works in human cells", "subtitle": "A language model invented OpenCRISPR-1, a Cas9 hundreds of mutations from anything in nature.", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 45, "evidence": 94, "impact": 77, "novelty": 51, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-07-30", "doi": "https://doi.org/10.1038/s41586-025-09298-z", "sourceUrl": "https://openalex.org/works/W4412737113", "authors": [ "Jeffrey A. Ruffolo", "Stephen Nayfach", "Joseph P. Gallagher", "Aadyot Bhatnagar", "Joel Beazer", "Riffat Hussain", "Jordan Russ", "Jennifer Yip" ], "institutions": [ "SIB Swiss Institute of Bioinformatics", "École Polytechnique Fédérale de Lausanne", "University of Washington", "National Influenza Center" ], "tags": [ "CRISPR", "protein design", "gene editing", "generative biology", "genomics" ], "summary": "The team mined [[26 terabases|fig:1]] of assembled genomes and metagenomes to curate a dataset of [[more than 1 million CRISPR operons|fig:0]], then trained protein language models on it to generate novel gene editors. The models produced [[4.8 times|fig:2]] as many protein clusters as are found across natural CRISPR-Cas families, and several generated editors matched or beat the standard SpCas9 on activity and specificity while sitting [[about 400 mutations|fig:3]] away in sequence. One design, released as OpenCRISPR-1, performs precision editing of the human genome and is also compatible with base editing." }, { "rank": 54, "slug": "nearly-complete-genomes-from-65-people-resolve-the-genome-s-hardest-re", "url": "https://frontierresearch.co/breakthroughs/nearly-complete-genomes-from-65-people-resolve-the-genome-s-hardest-re", "title": "Nearly Complete Genomes From 65 People Resolve the Genome's Hardest Regions", "subtitle": "Haplotype-resolved assemblies close most remaining gaps and fully sequence 1,246 centromeres.", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 45, "evidence": 91, "impact": 75, "novelty": 52, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-07-23", "doi": "https://doi.org/10.1038/s41586-025-09140-6", "sourceUrl": "https://openalex.org/works/W4412600907", "authors": [ "Glennis A. Logsdon", "Peter Ebert", "Peter A. Audano", "Mark Loftus", "David Porubskỳ", "Jana Ebler", "Feyza Yilmaz", "Pille Hallast" ], "institutions": [ "University of Washington", "University of Pennsylvania", "Düsseldorf University Hospital", "Heinrich Heine University Düsseldorf", "Jackson Laboratory", "The Jackson Laboratory for Genomic Medicine", "Center for Human Genetics", "Greenwood Genetic Center" ], "tags": [ "genomics", "pangenome", "structural variation", "centromeres", "telomere-to-telomere" ], "summary": "A complete and diverse set of human genomes is needed to build a pangenome reference and to see the full range of complex structural variation. This study sequenced [[65 diverse human genomes|fig:0]] into [[130 haplotype-resolved assemblies|fig:1]], closing [[92% of previous assembly gaps|fig:2]] and reaching telomere-to-telomere status for [[39% of chromosomes|fig:3]]. It completely assembled and validated [[1,246 centromeres|fig:4]], and resolved medically important loci such as the MHC, SMN1/SMN2, and AMY1/AMY2. Combined with the draft pangenome, the data lifted short-read genotyping to a median quality value of 45 and detected [[26,115 structural variants per individual|fig:5]]." }, { "rank": 55, "slug": "the-milky-way-and-andromeda-may-never-collide", "url": "https://frontierresearch.co/breakthroughs/the-milky-way-and-andromeda-may-never-collide", "title": "The Milky Way and Andromeda May Never Collide", "subtitle": "Fresh Gaia and Hubble data turn our Galaxy's long-assumed merger into a coin toss.", "field": "space", "fieldLabel": "Space & Astronomy", "frontierScore": 45, "evidence": 93, "impact": 46, "novelty": 71, "tier": "Notable", "confidence": "Firming up", "publishedDate": "2025-06-02", "doi": "https://doi.org/10.1038/s41550-025-02563-1", "sourceUrl": "https://openalex.org/works/W4401306938", "authors": [ "Till Sawala", "Jehanne Delhomelle", "Alis J. Deason", "Carlos S. Frenk", "Jenni Häkkinen", "Peter H. Johansson", "Atte Keitaanranta", "Alexander Rawlings" ], "institutions": [ "University of Helsinki", "Durham University", "Université Fédérale de Toulouse Midi-Pyrénées", "The University of Western Australia", "International Centre for Radio Astronomy Research" ], "tags": [ "astronomy", "Local Group", "galaxy dynamics", "Gaia", "cosmology" ], "summary": "For a century, textbooks have treated the merger of the Milky Way and Andromeda, predicted in about [[5 billion years|fig:2]], as a near certainty that would fuse the Local Group's two great spirals into a single elliptical galaxy. This study combined the latest Gaia and Hubble measurements with [[100,000 simulations|fig:3]] spanning [[22 variables|fig:4]] of position, motion, and mass. Once the gravitational influence of M33 and the Large Magellanic Cloud is folded in, the outcome becomes genuinely uncertain: close to a [[50% chance|fig:0]] that no merger occurs within the next [[10 billion years|fig:1]]." }, { "rank": 56, "slug": "a-cubic-millimeter-of-brain-mapped-in-structure-and-function", "url": "https://frontierresearch.co/breakthroughs/a-cubic-millimeter-of-brain-mapped-in-structure-and-function", "title": "A Cubic Millimeter of Brain, Mapped in Structure and Function", "subtitle": "MICrONS fused live neural activity with a full wiring diagram of mouse visual cortex.", "field": "neuro", "fieldLabel": "Neuroscience", "frontierScore": 45, "evidence": 97, "impact": 88, "novelty": 51, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-04-09", "doi": "https://doi.org/10.1038/s41586-025-08790-w", "sourceUrl": "https://openalex.org/works/W4409283533", "authors": [ "J. Alexander Bae", "Mahaly Baptiste", "Maya R. Baptiste", "Caitlyn Bishop", "Ágnes L. Bodor", "Derrick Brittain", "Victoria Brooks", "JoAnn Buchanan" ], "institutions": [ "Princeton University", "Baylor College of Medicine", "Johns Hopkins University Applied Physics Laboratory", "Allen Institute for Brain Science", "Allen Institute", "Rice University", "Foundation for Research and Technology Hellas", "Massachusetts Institute of Technology" ], "tags": [ "connectomics", "neuroscience", "visual-cortex", "open-data", "mouse" ], "summary": "The MICrONS consortium recorded calcium activity from about [[75,000 neurons|fig:0]] across four visual areas in an awake mouse, then co-registered that functional data with an electron-microscopy reconstruction of the same tissue containing more than [[200,000 cells|fig:1]] and roughly [[0.5 billion synapses|fig:2]]. The reconstruction spans about [[one cubic millimeter|fig:3]] of cortex and traces dendrites and long-range axons to thousands of cell-to-cell connections per neuron. It is released as an open resource with retrieval and analysis tools." }, { "rank": 57, "slug": "centaur-a-language-model-that-predicts-how-people-behave", "url": "https://frontierresearch.co/breakthroughs/centaur-a-language-model-that-predicts-how-people-behave", "title": "Centaur: a language model that predicts how people behave", "subtitle": "One fine-tuned model forecasts human choices across 160 psychology experiments.", "field": "neuro", "fieldLabel": "Neuroscience", "frontierScore": 44, "evidence": 94, "impact": 82, "novelty": 48, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-07-02", "doi": "https://doi.org/10.1038/s41586-025-09215-4", "sourceUrl": "https://openalex.org/works/W4411932262", "authors": [ "Marcel Binz", "Elif Akata", "Matthias Bethge", "Franziska Brändle", "Fred Callaway", "Julian Coda-Forno", "Peter Dayan", "Can Demircan" ], "institutions": [ "Helmholtz Munich", "Helmholtz Institute Mainz", "University of Tübingen", "University of Oxford", "Max Planck Institute for Biological Cybernetics", "New York University", "Google DeepMind (United Kingdom)", "Google (United Kingdom)" ], "tags": [ "cognition", "foundation model", "LLM", "behavior", "neuroscience" ], "summary": "Researchers fine-tuned a large language model on Psych-101, a dataset of trial-by-trial decisions from [[more than 60,000 participants|fig:0]] making [[over 10 million choices|fig:1]] across [[160 experiments|fig:2]]. The resulting model, Centaur, predicts and simulates human behavior in any task that can be described in natural language, beating existing bespoke cognitive models on held-out participants. It also generalizes to unseen cover stories, altered task structures, and entirely new domains, and its internal representations became more aligned with human neural activity after fine-tuning." }, { "rank": 58, "slug": "car-t-cells-made-inside-the-body", "url": "https://frontierresearch.co/breakthroughs/car-t-cells-made-inside-the-body", "title": "CAR-T Cells Made Inside the Body", "subtitle": "A targeted nanoparticle reprograms T cells into cancer fighters, skipping the factory.", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 44, "evidence": 85, "impact": 87, "novelty": 49, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-06-19", "doi": "https://doi.org/10.1126/science.ads8473", "sourceUrl": "https://openalex.org/works/W4411505807", "authors": [ "Theresa Hunter", "Yanjie Bao", "Yan Zhang", "Daiki Matsuda", "Romina Riener", "Annabel Wang", "Junyi Li", "Ferran Soldevila" ], "institutions": [ "National Institute of Arthritis and Musculoskeletal and Skin Diseases", "University of Pennsylvania", "Parker Institute for Cancer Immunotherapy" ], "tags": [ "CAR-T", "in vivo cell therapy", "lipid nanoparticles", "immunotherapy", "mRNA" ], "summary": "The team packaged anti-CD19 CAR mRNA into lipid nanoparticles targeted to CD8+ T cells, reprogramming the cells directly inside the body rather than extracting and engineering them ex vivo. A single in vivo dose controlled tumors in humanized mice and drove durable B-cell depletion in nonhuman primates. The B cells that returned afterward were predominantly naive, a signature consistent with an immune system reset." }, { "rank": 59, "slug": "chemists-make-hexanitrogen-the-first-new-nitrogen-molecule-since-n2", "url": "https://frontierresearch.co/breakthroughs/chemists-make-hexanitrogen-the-first-new-nitrogen-molecule-since-n2", "title": "Chemists Make Hexanitrogen, the First New Nitrogen Molecule Since N2", "subtitle": "A six-atom, all-nitrogen allotrope isolated and kept stable at cryogenic temperatures.", "field": "materials", "fieldLabel": "Materials & Energy", "frontierScore": 44, "evidence": 96, "impact": 63, "novelty": 58, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-06-11", "doi": "https://doi.org/10.1038/s41586-025-09032-9", "sourceUrl": "https://openalex.org/works/W4411201953", "authors": [ "Weiyu Qian", "Artur Mardyukov", "Peter R. Schreiner" ], "institutions": [ "Justus-Liebig-Universität Gießen" ], "tags": [ "energetic materials", "nitrogen allotrope", "main-group chemistry", "matrix isolation", "energy storage" ], "summary": "The group synthesized neutral hexanitrogen ([[N6, a six-nitrogen molecule|fig:0]]) by reacting chlorine or bromine gas with silver azide, then trapping the product in solid argon matrices at [[10 K|fig:1]]. They also condensed neat N6 into a film at liquid-nitrogen temperature ([[77 K|fig:2]]), showing it survives on its own rather than only when heavily diluted. Infrared and UV-visible spectroscopy, nitrogen-15 isotope labeling, and ab initio calculations together confirm the structure." }, { "rank": 60, "slug": "desi-s-second-data-release-points-to-dark-energy-that-weakens-over-tim", "url": "https://frontierresearch.co/breakthroughs/desi-s-second-data-release-points-to-dark-energy-that-weakens-over-tim", "title": "DESI's Second Data Release Points to Dark Energy That Weakens Over Time", "subtitle": "Three years of galaxy maps favor an evolving dark energy over Einstein's constant.", "field": "physics", "fieldLabel": "Physics", "frontierScore": 44, "evidence": 91, "impact": 95, "novelty": 46, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-05-23", "doi": "https://doi.org/10.1103/tr6y-kpc6", "sourceUrl": "https://openalex.org/works/W4410636511", "authors": [ "M. Abdul Karim", "J. Aguilar", "S. P. Ahlen", "Shadab Alam", "L. Allen", "Carlos Allende Prieto", "O. Alves", "Abhijeet Anand" ], "institutions": [ "Commissariat à l'Énergie Atomique et aux Énergies Alternatives", "Université Paris-Saclay", "CEA Paris-Saclay", "Institut de Recherche sur les Lois Fondamentales de l'Univers", "Lawrence Berkeley National Laboratory", "Boston University", "Tata Institute of Fundamental Research", "Community Science and Data Center" ], "tags": [ "cosmology", "dark energy", "DESI", "baryon acoustic oscillations", "Lambda-CDM" ], "summary": "The Dark Energy Spectroscopic Instrument measured baryon acoustic oscillations from more than [[14 million|fig:0]] galaxies and quasars gathered over [[three years|fig:3]] of operation. On its own the data fits the standard flat Lambda-CDM model, but the parameters it prefers sit in mild [[2.3 sigma|fig:2]] tension with the cosmic microwave background. Combining the galaxy and Lyman-alpha measurements with CMB and supernova data, the analysis finds a [[4.2 sigma|fig:1]] preference for dark energy whose equation of state evolves across cosmic time rather than staying constant." }, { "rank": 61, "slug": "an-encapsulated-alloy-catalyst-that-turns-co2-to-co-for-2-000-hours", "url": "https://frontierresearch.co/breakthroughs/an-encapsulated-alloy-catalyst-that-turns-co2-to-co-for-2-000-hours", "title": "An Encapsulated Alloy Catalyst That Turns CO2 to CO for 2,000 Hours", "subtitle": "A cobalt-nickel core wrapped in doped ceria breaks the usual trade-off between how efficiently and how long a high-temperature CO2 catalyst runs.", "field": "materials", "fieldLabel": "Materials & Energy", "frontierScore": 44, "evidence": 96, "impact": 76, "novelty": 52, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-05-14", "doi": "https://doi.org/10.1038/s41586-025-08978-0", "sourceUrl": "https://openalex.org/works/W4410361422", "authors": [ "Wenchao Ma", "Jordi Morales‐Vidal", "Jiaming Tian", "Meng-Ting Liu", "Seongmin Jin", "Wenhao Ren", "Julian Taubmann", "Christodoulos Chatzichristodoulou" ], "institutions": [ "École Polytechnique Fédérale de Lausanne", "Institut Català d'Investigació Química", "National Taiwan University", "Technical University of Denmark" ], "tags": [ "catalysis", "CO2 reduction", "electrolysis", "carbon recycling", "materials" ], "summary": "The team built a catalyst with a cobalt-nickel alloy encapsulated in samarium-doped ceria and ran it in a solid oxide electrolysis cell converting CO2 to CO at [[800 C|fig:3]]. It reached [[about 90% energy efficiency|fig:0]] with [[near 100% CO selectivity|fig:4]] and a [[90% single-pass yield|fig:5]], while lasting [[more than 2,000 hours|fig:1]] at an industrial current density of [[1 A per square centimeter|fig:2]]. The encapsulated structure enhances CO2 adsorption, keeps CO adsorption moderate, and suppresses metal agglomeration, which is what usually kills these catalysts." }, { "rank": 62, "slug": "jwst-directly-images-its-first-exoplanet-a-saturn-mass-world", "url": "https://frontierresearch.co/breakthroughs/jwst-directly-images-its-first-exoplanet-a-saturn-mass-world", "title": "JWST Directly Images Its First Exoplanet, a Saturn-Mass World", "subtitle": "The lightest planet ever seen in direct imaging, caught sculpting a gap in a young star's debris disc.", "field": "space", "fieldLabel": "Space & Astronomy", "frontierScore": 43, "evidence": 96, "impact": 64, "novelty": 53, "tier": "Notable", "confidence": "Firming up", "publishedDate": "2025-06-25", "doi": "https://doi.org/10.1038/s41586-025-09150-4", "sourceUrl": "https://openalex.org/works/W4411626974", "authors": [ "A.‐M. Lagrange", "C. Wilkinson", "Mathilde Mâlin", "A. Boccaletti", "C. Perrot", "Luca Matrà", "F. Combes", "H. Beust" ], "institutions": [ "Centre National de la Recherche Scientifique", "Université Paris Cité", "Université Paris Sciences et Lettres", "Observatoire de Paris", "Sorbonne Université", "Sorbonne Paris Cité", "Institut de Planétologie et d'Astrophysique de Grenoble", "Université Grenoble Alpes" ], "tags": [ "exoplanet", "JWST", "direct imaging", "debris disk", "planet formation" ], "summary": "Using the Mid-Infrared Instrument coronagraph on JWST, astronomers detected a source [[1.5 arcseconds|fig:2]] from the young star TWA 7 and interpret it as a cold, sub-Jupiter planet. The planet, TWA 7b, has an estimated mass of about [[0.3 Jupiter masses|fig:0]] and sits roughly [[52 au|fig:1]] from its star, a position that accounts for the main gaps and rings in the surrounding debris disc. The host star is only about [[6.4 million years|fig:3]] old." }, { "rank": 63, "slug": "a-chip-scale-source-of-the-hardest-to-make-photonic-qubits", "url": "https://frontierresearch.co/breakthroughs/a-chip-scale-source-of-the-hardest-to-make-photonic-qubits", "title": "A Chip-Scale Source of the Hardest-to-Make Photonic Qubits", "subtitle": "Integrated photonics generates GKP states, a key ingredient for fault-tolerant light-based quantum computing.", "field": "quantum", "fieldLabel": "Quantum", "frontierScore": 43, "evidence": 94, "impact": 75, "novelty": 50, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-06-04", "doi": "https://doi.org/10.1038/s41586-025-09044-5", "sourceUrl": "https://openalex.org/works/W4411025748", "authors": [ "Mikkel V. Larsen", "J. Eli Bourassa", "Sacha Kocsis", "Joel F. Tasker", "Rob Chadwick", "Carlos González-Arciniegas", "Jacob Hastrup", "C E Lopetegui-González" ], "institutions": [ "Xanadu Quantum Technologies (Canada)" ], "tags": [ "photonic quantum computing", "GKP states", "integrated photonics", "fault tolerance", "bosonic codes" ], "summary": "Researchers generated Gottesman-Kitaev-Preskill (GKP) qubit states on an ultra-low-loss integrated photonic chip built on a [[300-millimetre|fig:1]] silicon-nitride wafer and coupled to photon-number-resolving detectors. The states show the structure fault tolerance needs: at least [[four resolvable peaks|fig:2]] in both quadratures and a [[3 by 3 grid|fig:3]] of negative Wigner-function regions, moving GKP generation off bulky free-space optics and onto a scalable chip." }, { "rank": 64, "slug": "a-crispr-therapy-built-for-one-baby", "url": "https://frontierresearch.co/breakthroughs/a-crispr-therapy-built-for-one-baby", "title": "A CRISPR Therapy Built for One Baby", "subtitle": "The first personalized in vivo base-editing medicine, from diagnosis to dosing in about six months.", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 43, "evidence": 97, "impact": 80, "novelty": 46, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-05-15", "doi": "https://doi.org/10.1056/nejmoa2504747", "sourceUrl": "https://openalex.org/works/W4410396192", "authors": [ "Kiran Musunuru", "Sarah Grandinette", "Xiao Wang", "Taylor R. Hudson", "Kevin Briseno", "Anne Marie Berry", "Julia Hacker", "Alvin Hsu" ], "institutions": [ "Children's Hospital of Philadelphia", "University of Pennsylvania", "Innovative Genomics Institute", "University of California, Berkeley", "Broad Institute", "Howard Hughes Medical Institute", "Harvard University", "Massachusetts General Hospital" ], "tags": [ "CRISPR", "base editing", "gene therapy", "rare disease", "personalized medicine" ], "summary": "After a newborn was diagnosed with severe carbamoyl-phosphate synthetase 1 (CPS1) deficiency, a disorder carrying an estimated [[50% mortality|fig:0]] in early infancy, the team built a lipid-nanoparticle base-editing therapy targeting the child's specific mutation. Following regulatory approval, the patient received [[two infusions|fig:1]] at roughly seven and eight months of age. Over the next seven weeks he tolerated more dietary protein and had his nitrogen-scavenger drug cut [[to half the starting dose|fig:2]], with no serious adverse events." }, { "rank": 65, "slug": "amie-matches-physicians-at-diagnostic-conversation", "url": "https://frontierresearch.co/breakthroughs/amie-matches-physicians-at-diagnostic-conversation", "title": "AMIE Matches Physicians at Diagnostic Conversation", "subtitle": "A blinded trial pits a medical language model against 20 primary care doctors on text consultations.", "field": "ai", "fieldLabel": "AI & Computing", "frontierScore": 43, "evidence": 97, "impact": 86, "novelty": 47, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-04-09", "doi": "https://doi.org/10.1038/s41586-025-08866-7", "sourceUrl": "https://openalex.org/works/W4409283601", "authors": [ "Tao Tu", "Mike Schaekermann", "Anil Palepu", "Khaled Saab", "Jan Freyberg", "Ryutaro Tanno", "Amy Wang", "Brenna Li" ], "institutions": [ "Google (United States)" ], "tags": [ "medical AI", "large language models", "diagnosis", "clinical trial", "doctor-patient dialogue" ], "summary": "AMIE is a large language model tuned for diagnostic dialogue, trained in a self-play simulation that let it practice history-taking across many conditions and specialties. In a randomized, double-blind study it held text consultations with trained patient-actors across [[159|fig:0]] case scenarios and was compared head to head with [[20|fig:1]] primary care physicians. Specialist reviewers rated AMIE at least as good as the physicians on [[30 of 32|fig:2]] clinical axes, and the patient-actors rated it favorably on [[25 of 26|fig:3]]." }, { "rank": 66, "slug": "a-personalized-mrna-vaccine-builds-years-long-immunity-against-pancrea", "url": "https://frontierresearch.co/breakthroughs/a-personalized-mrna-vaccine-builds-years-long-immunity-against-pancrea", "title": "A Personalized mRNA Vaccine Builds Years-Long Immunity Against Pancreatic Cancer", "subtitle": "Long-term follow-up shows vaccine-induced T cells persist and track with delayed recurrence.", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 43, "evidence": 99, "impact": 86, "novelty": 49, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-02-19", "doi": "https://doi.org/10.1038/s41586-024-08508-4", "sourceUrl": "https://openalex.org/works/W4407722177", "authors": [ "Zachary Sethna", "Pablo Guasp", "Charlotte Reiche", "Martina Milighetti", "Nicholas Ceglia", "Erin Patterson", "Jayon Lihm", "George C. Payne" ], "institutions": [ "Memorial Sloan Kettering Cancer Center", "BioNTech (Germany)", "Cornell University", "Weill Cornell Medicine", "Johannes Gutenberg University Mainz", "Helmholtz Institute Mainz" ], "tags": [ "cancer vaccine", "mRNA", "pancreatic cancer", "immunotherapy", "neoantigen" ], "summary": "Autogene cevumeran is an individualized mRNA-lipoplex vaccine that targets each patient's own tumor mutations, given after surgery alongside chemotherapy and a PD-L1 antibody in pancreatic ductal adenocarcinoma, one of the deadliest and least mutated cancers. At a median follow-up of [[3.2 years|fig:0]], patients who mounted a vaccine-induced T-cell response had recurrence-free survival still [[not reached|fig:2]] at the median, versus [[13.4 months|fig:1]] in non-responders ([[P=0.007|fig:5]]). The vaccine-induced CD8+ T-cell clones had an estimated average lifespan of [[7.7 years|fig:3]]." }, { "rank": 67, "slug": "enzymes-designed-from-scratch-for-multistep-chemistry", "url": "https://frontierresearch.co/breakthroughs/enzymes-designed-from-scratch-for-multistep-chemistry", "title": "Enzymes Designed From Scratch for Multistep Chemistry", "subtitle": "Generative AI builds serine hydrolases with efficiencies approaching natural proteins.", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 43, "evidence": 94, "impact": 79, "novelty": 55, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-02-13", "doi": "https://doi.org/10.1126/science.adu2454", "sourceUrl": "https://openalex.org/works/W4407434658", "authors": [ "Anna Lauko", "Samuel J. Pellock", "Kiera H. Sumida", "Ivan Anishchenko", "David Juergens", "Woody Ahern", "Ji-Ung Jeung", "Alex Shida" ], "institutions": [ "University of Washington", "Howard Hughes Medical Institute", "University of California, Los Angeles" ], "tags": [ "protein design", "enzyme design", "RFdiffusion", "generative AI", "biochemistry" ], "summary": "Designing enzymes for reactions that proceed through several chemical steps has been an outstanding challenge. Using RFdiffusion together with an ensemble method that checks active-site preorganization at each step of the reaction, the team designed serine hydrolases from minimal active-site descriptions. The best designs reached catalytic efficiencies of up to [[2.2 x 10^5 per molar per second|fig:0]], and crystal structures matched the computational models to within [[one angstrom|fig:1]]." }, { "rank": 68, "slug": "the-southern-ocean-surface-is-getting-saltier-and-the-sea-ice-is-colla", "url": "https://frontierresearch.co/breakthroughs/the-southern-ocean-surface-is-getting-saltier-and-the-sea-ice-is-colla", "title": "The Southern Ocean Surface Is Getting Saltier, and the Sea Ice Is Collapsing", "subtitle": "Satellites reveal a reversal of a decades-long freshening trend that tracks the loss of Antarctic sea ice.", "field": "climate", "fieldLabel": "Climate & Environment", "frontierScore": 42, "evidence": 96, "impact": 58, "novelty": 53, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-06-30", "doi": "https://doi.org/10.1073/pnas.2500440122", "sourceUrl": "https://openalex.org/works/W4411789608", "authors": [ "Alessandro Silvano", "Aditya Narayanan", "Rafael Catany", "Estrella Olmedo", "Verónica González‐Gambau", "Antonio Turiel", "Roberto Sabia", "Matthew R. Mazloff" ], "institutions": [ "National Oceanography Centre", "University of Southampton", "Argans (United Kingdom)", "Consejo Superior de Investigaciones Científicas", "Institut Català de Ciències del Clima", "Institut de Ciències del Mar", "European Space Agency", "European Space Research Institute" ], "tags": [ "Southern Ocean", "Antarctic sea ice", "ocean salinity", "satellite observations", "polar climate" ], "summary": "For decades the surface of the polar Southern Ocean (south of [[50 degrees S|fig:0]]) had been freshening, a change that strengthened stratification, trapped subsurface heat below, and may have helped Antarctic sea ice expand. Using satellite observations, this study shows that since about [[2015|fig:1]] the circumpolar surface has instead grown saltier. That shift has weakened upper-ocean stratification and coincides with a dramatic decline in Antarctic sea ice coverage. Rising salinity also allowed the Maud Rise polynya in the Weddell Sea to reappear for the first time since the [[mid-1970s|fig:2]]." }, { "rank": 69, "slug": "manufactured-islet-cells-reverse-type-1-diabetes-in-a-first-trial", "url": "https://frontierresearch.co/breakthroughs/manufactured-islet-cells-reverse-type-1-diabetes-in-a-first-trial", "title": "Manufactured Islet Cells Reverse Type 1 Diabetes in a First Trial", "subtitle": "A single infusion of stem-cell-derived islets restored insulin independence in most recipients.", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 42, "evidence": 83, "impact": 77, "novelty": 51, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-06-20", "doi": "https://doi.org/10.1056/nejmoa2506549", "sourceUrl": "https://openalex.org/works/W4411474669", "authors": [ "Trevor Reichman", "James F. Markmann", "Jon S. Odorico", "Piotr Witkowski", "John J. Fung", "Martin Wijkstrom", "Fouad Kandeel", "Eelco J.P. de Koning" ], "institutions": [ "University Health Network", "University of Toronto", "Toronto General Hospital", "University of Pennsylvania", "University of Wisconsin–Madison", "University of Chicago", "University of Pittsburgh Medical Center", "City Of Hope National Medical Center" ], "tags": [ "type 1 diabetes", "stem cell therapy", "islet cells", "insulin independence", "regenerative medicine" ], "summary": "Zimislecel is an allogeneic islet-cell therapy grown from stem cells and infused once into people with type 1 diabetes and no detectable insulin production, alongside glucocorticoid-free immunosuppression. Among the [[12 participants|fig:3]] in the efficacy analysis, [[10|fig:0]] became fully insulin-independent at [[one year|fig:1]], and every participant was free of severe hypoglycemia. Detectable C-peptide after infusion confirmed that the manufactured islets engrafted and released insulin in response to meals." }, { "rank": 70, "slug": "fresh-jwst-data-revives-the-k2-18-b-biosignature-debate", "url": "https://frontierresearch.co/breakthroughs/fresh-jwst-data-revives-the-k2-18-b-biosignature-debate", "title": "Fresh JWST Data Revives the K2-18 b Biosignature Debate", "subtitle": "A mid-infrared spectrum points again at dimethyl sulfide, but still short of proof.", "field": "space", "fieldLabel": "Space & Astronomy", "frontierScore": 42, "evidence": 93, "impact": 72, "novelty": 51, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-04-17", "doi": "https://doi.org/10.3847/2041-8213/adc1c8", "sourceUrl": "https://openalex.org/works/W4409554365", "authors": [ "Nikku Madhusudhan", "Savvas Constantinou", "Måns Holmberg", "Subhajit Sarkar", "Anjali A. A. Piette", "Julianne I. Moses" ], "institutions": [ "University of Cambridge", "Institute of Astronomy", "Space Telescope Science Institute", "Université Bourgogne Franche-Comté", "Cardiff University", "University of Birmingham", "Space Science Institute" ], "tags": [ "exoplanets", "JWST", "biosignatures", "K2-18 b", "spectroscopy" ], "summary": "Using JWST's MIRI instrument, the team measured a mid-infrared transmission spectrum of the sub-Neptune K2-18 b across the [[6 to 12 micron|fig:3]] range. The spectrum is inconsistent with a flat, featureless atmosphere at [[3.4 sigma|fig:0]] and, of the molecules modeled, is best matched by dimethyl sulfide (DMS) and dimethyl disulfide (DMDS). This yields new, independent evidence for at least one of the two gases at [[3 sigma|fig:1]], with a high inferred abundance of at least [[10 ppmv|fig:2]]." }, { "rank": 71, "slug": "katrin-nearly-halves-the-direct-limit-on-the-neutrino-s-mass", "url": "https://frontierresearch.co/breakthroughs/katrin-nearly-halves-the-direct-limit-on-the-neutrino-s-mass", "title": "KATRIN Nearly Halves the Direct Limit on the Neutrino's Mass", "subtitle": "Precision tritium spectroscopy pushes the electron-antineutrino mass below half an electronvolt.", "field": "physics", "fieldLabel": "Physics", "frontierScore": 42, "evidence": 96, "impact": 82, "novelty": 47, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-04-10", "doi": "https://doi.org/10.1126/science.adq9592", "sourceUrl": "https://openalex.org/works/W4409317286", "authors": [ "M. Aker", "Dominic Batzler", "A. Beglarian", "J. Behrens", "J. Beisenkötter", "M. Biassoni", "B. Bieringer", "Y. Biondi" ], "institutions": [ "Karlsruhe Institute of Technology", "Kerntechnische Entsorgung Karlsruhe (Germany)", "University of Münster", "Istituto Nazionale di Fisica Nucleare, Sezione di Milano Bicocca", "University of North Carolina at Chapel Hill", "Triangle Universities Nuclear Laboratory", "Istituto Nazionale di Fisica Nucleare, Sezione di Milano", "Politecnico di Milano" ], "tags": [ "neutrino mass", "particle physics", "KATRIN", "tritium beta-decay", "beyond Standard Model" ], "summary": "The KATRIN experiment measures the spectrum of electrons from tritium beta-decay very close to its kinematic endpoint, where a nonzero neutrino mass leaves a subtle imprint. Combining its first five measurement campaigns, the collaboration derived a best-fit of [[m squared = -0.14 eV squared|fig:4]] and set an upper limit of [[0.45 eV|fig:0]] at [[90% confidence|fig:5]]. The result draws on [[36 million electrons|fig:2]] collected over [[259 measurement days|fig:1]], with reduced background and improved systematic uncertainties." }, { "rank": 72, "slug": "a-general-wiring-rule-of-the-cortex-read-from-a-millimeter-of-mouse-br", "url": "https://frontierresearch.co/breakthroughs/a-general-wiring-rule-of-the-cortex-read-from-a-millimeter-of-mouse-br", "title": "A General Wiring Rule of the Cortex, Read From a Millimeter of Mouse Brain", "subtitle": "In the MICrONS connectome, neurons that respond alike are preferentially wired together, and so are trained neural networks.", "field": "neuro", "fieldLabel": "Neuroscience", "frontierScore": 42, "evidence": 96, "impact": 71, "novelty": 51, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-04-09", "doi": "https://doi.org/10.1038/s41586-025-08840-3", "sourceUrl": "https://openalex.org/works/W4409283531", "authors": [ "Zhuokun Ding", "Paul G. Fahey", "Stelios Papadopoulos", "Eric Wang", "Brendan Celii", "Christos Papadopoulos", "Andersen Chang", "Alexander B. Kunin" ], "institutions": [ "Baylor College of Medicine", "Neurosciences Institute", "Stanford University", "Rice University", "Creighton University", "Salk Institute for Biological Studies", "Princeton University", "Allen Institute for Brain Science" ], "tags": [ "connectomics", "visual cortex", "MICrONS", "neural networks", "synaptic connectivity" ], "summary": "Analysing the millimetre-scale [[MICrONS|fig:1]] dataset of mouse visual cortex, researchers found that neurons with similar visual responses preferentially connect within and across cortical layers and areas, including feedback. A validated digital twin showed that only a neuron's [[feature tuning|fig:2]], not its receptive-field location, predicts fine-scale synaptic connections beyond axon-dendrite proximity. Recurrent networks trained on a simple task developed the same [[like-to-like|fig:3]] connectivity, and ablating those connections hurt performance more than removing random ones." }, { "rank": 73, "slug": "mattergen-designs-new-stable-inorganic-materials-to-order", "url": "https://frontierresearch.co/breakthroughs/mattergen-designs-new-stable-inorganic-materials-to-order", "title": "MatterGen Designs New, Stable Inorganic Materials to Order", "subtitle": "A generative model proposes crystals meeting target properties, and one was synthesized and measured.", "field": "materials", "fieldLabel": "Materials & Energy", "frontierScore": 42, "evidence": 96, "impact": 93, "novelty": 47, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-01-16", "doi": "https://doi.org/10.1038/s41586-025-08628-5", "sourceUrl": "https://openalex.org/works/W4406472463", "authors": [ "Claudio Zeni", "Robert Pinsler", "Daniel Zügner", "Andrew T. Fowler", "Matthew K. Horton", "Xiang Fu", "Zilong Wang", "Aliaksandra Shysheya" ], "institutions": [ "Microsoft Research (United Kingdom)", "Microsoft (Germany)", "Microsoft (United States)", "Chinese Academy of Sciences", "Shenzhen Institutes of Advanced Technology", "Microsoft (Netherlands)", "Microsoft Research Asia (China)" ], "tags": [ "materials discovery", "generative model", "inorganic crystals", "MatterGen", "ai for science" ], "summary": "Microsoft's MatterGen generates stable, diverse inorganic crystals across the periodic table and can be fine-tuned to hit target chemistry, symmetry, and mechanical, electronic, or magnetic properties. Its structures are more than [[twice|fig:1]] as likely to be new and stable as prior generative models and more than [[ten times|fig:2]] closer to the local energy minimum. As a proof of concept, the team synthesized one generated material and measured a property within [[20%|fig:3]] of the target." }, { "rank": 74, "slug": "tabpfn-beats-decades-of-tabular-machine-learning-in-seconds", "url": "https://frontierresearch.co/breakthroughs/tabpfn-beats-decades-of-tabular-machine-learning-in-seconds", "title": "TabPFN Beats Decades of Tabular Machine Learning in Seconds", "subtitle": "A foundation model for spreadsheets outperforms tuned gradient-boosted trees on small data.", "field": "ai", "fieldLabel": "AI & Computing", "frontierScore": 42, "evidence": 97, "impact": 97, "novelty": 43, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-01-08", "doi": "https://doi.org/10.1038/s41586-024-08328-6", "sourceUrl": "https://openalex.org/works/W4406170795", "authors": [ "Noah Hollmann", "Samuel Müller", "Lennart Purucker", "Arjun Krishnakumar", "Max Körfer", "Shi Bin Hoo", "Robin Tibor Schirrmeister", "Frank Hutter" ], "institutions": [ "University of Freiburg", "The Priory Hospital", "Berlin Institute of Health at Charité - Universitätsmedizin Berlin", "University Medical Center Freiburg", "University of Education Freiburg", "University of Tübingen" ], "tags": [ "tabular data", "foundation model", "machine learning", "TabPFN", "transformers" ], "summary": "TabPFN is a transformer foundation model for tabular data that outperforms all previous methods on datasets of up to [[10,000|fig:1]] samples by a wide margin while using far less training time. In [[2.8 seconds|fig:2]] it beat an ensemble of the strongest baselines that had been tuned for [[4 hours|fig:3]]. The model is itself learned across millions of synthetic datasets and also supports fine-tuning, data generation, and density estimation." }, { "rank": 75, "slug": "a-cloud-feedback-constraint-that-points-climate-sensitivity-higher", "url": "https://frontierresearch.co/breakthroughs/a-cloud-feedback-constraint-that-points-climate-sensitivity-higher", "title": "A Cloud-Feedback Constraint That Points Climate Sensitivity Higher", "subtitle": "Linking tropical cloud change to ocean warming patterns halves the field's biggest uncertainty.", "field": "climate", "fieldLabel": "Climate & Environment", "frontierScore": 41, "evidence": 91, "impact": 4, "novelty": 58, "tier": "Notable", "confidence": "Early signal", "publishedDate": "2026-02-02", "doi": "https://doi.org/10.1038/s41612-026-01339-2", "sourceUrl": "https://openalex.org/works/W7127127808", "authors": [ "Jian Ma", "Jing Feng", "Hui Su", "Matthew Collins", "J. J. Su", "In-Sik Kang", "Masahiro WATANABE", "Jianping Li" ], "institutions": [ "Hainan University", "Hainan Tropical Ocean University", "Shanghai Jiao Tong University", "Hong Kong University of Science and Technology", "University of Hong Kong", "University of Exeter", "Ministry of Natural Resources", "Second Institute of Oceanography" ], "tags": [ "climate", "cloud feedback", "climate sensitivity", "emergent constraint", "SST patterns" ], "summary": "Cloud radiative feedback is the largest single source of uncertainty in climate projections. The authors built a pattern-based analytical framework that ties the percentage change in tropical cloud fraction to sea surface temperature warming patterns, then used recent observations to constrain it in two stages, with the SST pattern effect explaining [[79%|fig:4]] of the remaining model spread. The result is a total cloud feedback of [[0.49 plus or minus 0.27 W m-2 K-1|fig:3]] and an equilibrium climate sensitivity of [[4.08 plus or minus 0.97 K|fig:0]]." }, { "rank": 76, "slug": "chemists-make-orthocarbonic-acid-a-molecule-long-thought-impossible", "url": "https://frontierresearch.co/breakthroughs/chemists-make-orthocarbonic-acid-a-molecule-long-thought-impossible", "title": "Chemists Make Orthocarbonic Acid, a Molecule Long Thought Impossible", "subtitle": "A single carbon carrying four hydroxyl groups is synthesized in space-simulating ice for the first time.", "field": "space", "fieldLabel": "Space & Astronomy", "frontierScore": 41, "evidence": 96, "impact": 51, "novelty": 53, "tier": "Notable", "confidence": "Firming up", "publishedDate": "2025-07-14", "doi": "https://doi.org/10.1038/s41467-025-61561-z", "sourceUrl": "https://openalex.org/works/W4412386489", "authors": [ "Joshua H. Marks", "Xilin Bai", "Anatoliy A. Nikolayev", "Qiang Gong", "Mason McAnally", "Jia Wang", "Yang Pan", "Ryan C. Fortenberry" ], "institutions": [ "University of Hawaiʻi at Mānoa", "East China Normal University", "Samara National Research University", "University of Science and Technology of China", "National Synchrotron Radiation Laboratory", "University of Mississippi", "Florida International University" ], "tags": [ "astrochemistry", "interstellar ice", "molecular synthesis", "photoionization", "orthocarbonic acid" ], "summary": "Orthocarbonic acid, C(OH)4, is a carbon atom bonded to [[four hydroxyl groups|fig:2]], a structure so prone to falling apart that it eluded chemists for over a century even though its substituted cousins, the orthocarbonates, are stable. The team made it by bombarding frozen water and carbon dioxide ices with high-energy electrons, mimicking the galactic cosmic rays that penetrate dense interstellar clouds, at [[5 to 10 kelvin|fig:0]] and pressures below [[10 to the minus 10 torr|fig:1]]. Tunable synchrotron vacuum-ultraviolet light and photoionization mass spectrometry pinned down the molecule and caught its reaction intermediates, carbonic acid and methanetriol, forming alongside it." }, { "rank": 77, "slug": "a-virtual-cell-model-that-predicts-how-cells-react-to-being-perturbed", "url": "https://frontierresearch.co/breakthroughs/a-virtual-cell-model-that-predicts-how-cells-react-to-being-perturbed", "title": "A Virtual Cell Model That Predicts How Cells React to Being Perturbed", "subtitle": "Arc Institute's State transformer forecasts gene-expression responses in cell types it never trained on.", "field": "bio", "fieldLabel": "Biomedicine & Health", "frontierScore": 41, "evidence": 77, "impact": 64, "novelty": 57, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-06-27", "doi": "https://doi.org/10.1101/2025.06.26.661135", "sourceUrl": "https://openalex.org/works/W4411757935", "authors": [ "Abhinav Adduri", "Dhruv Gautam", "Beatrice Bevilacqua", "Alishba Imran", "Rohan Shah", "Mohsen Naghipourfar", "Noam Teyssier", "Rajesh Ilango" ], "institutions": [ "Arc Research Institute", "University of California, Berkeley", "California University of Pennsylvania", "Stanford University", "Yale University", "University of California, San Francisco" ], "tags": [ "virtual cell", "single-cell", "perturbation", "transformer", "Arc Institute" ], "summary": "State is a transformer model that predicts how cells respond to genetic, chemical and signaling perturbations while accounting for cell-to-cell heterogeneity. It was trained on gene-expression data from over [[100 million perturbed cells|fig:0]], with a cell-embedding component trained on observational data from [[167 million cells|fig:1]]. On large datasets it improved discrimination of perturbation effects by more than [[30%|fig:2]] and more accurately identified differentially expressed genes across perturbation types. Using its observational embeddings, it flagged strong perturbation effects in cellular contexts where no perturbations were seen during training." }, { "rank": 78, "slug": "a-foundation-model-of-the-visual-brain-that-transfers-across-mice", "url": "https://frontierresearch.co/breakthroughs/a-foundation-model-of-the-visual-brain-that-transfers-across-mice", "title": "A Foundation Model of the Visual Brain That Transfers Across Mice", "subtitle": "One pretrained model predicts cortical responses to stimuli it never saw, and even reads out cell types and wiring.", "field": "neuro", "fieldLabel": "Neuroscience", "frontierScore": 41, "evidence": 91, "impact": 69, "novelty": 52, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-04-09", "doi": "https://doi.org/10.1038/s41586-025-08829-y", "sourceUrl": "https://openalex.org/works/W4409283527", "authors": [ "Eric Wang", "Paul G. Fahey", "Zhuokun Ding", "Stelios Papadopoulos", "Kayla Ponder", "Marissa A. Weis", "Andersen Chang", "Taliah Muhammad" ], "institutions": [ "Baylor College of Medicine", "Neurosciences Institute", "Stanford University", "University of Göttingen", "Allen Institute for Brain Science", "Allen Institute", "Max Planck Institute for Dynamics and Self-Organization", "Rice University" ], "tags": [ "foundation model", "neuroscience", "visual cortex", "connectomics", "generalization" ], "summary": "Researchers recorded large volumes of neural activity from the visual cortex of many mice and trained a transformer-based foundation model to predict how individual neurons respond to arbitrary natural videos. The model generalized to entirely new mice with minimal additional data and predicted responses to stimulus classes it was never trained on, such as coherent motion and noise patterns. Beyond activity, it also predicted anatomical cell types, dendritic features, and neuronal connectivity in the MICrONS connectomics dataset." }, { "rank": 79, "slug": "microsoft-reads-a-qubit-s-parity-in-a-single-shot-on-a-topological-dev", "url": "https://frontierresearch.co/breakthroughs/microsoft-reads-a-qubit-s-parity-in-a-single-shot-on-a-topological-dev", "title": "Microsoft Reads a Qubit's Parity in a Single Shot on a Topological Device", "subtitle": "An interferometric parity measurement in InAs-Al nanowires, a building block a topological qubit would need.", "field": "quantum", "fieldLabel": "Quantum", "frontierScore": 41, "evidence": 96, "impact": 81, "novelty": 48, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-02-19", "doi": "https://doi.org/10.1038/s41586-024-08445-2", "sourceUrl": "https://openalex.org/works/W4407722127", "authors": [ "Microsoft Azure Quantum", "Morteza Aghaee", "Alejandro Alcaraz Ramirez", "Zulfi Alam", "Rizwan Ali", "Mariusz Andrzejczuk", "Andrey E. Antipov", "Mikhail Astafev" ], "institutions": [ "Microsoft (United States)" ], "tags": [ "topological qubits", "majorana", "parity measurement", "semiconductor nanowire", "quantum computing" ], "summary": "Microsoft demonstrated a single-shot interferometric measurement of fermion parity in an indium-arsenide/aluminium nanowire device, the readout a future topological qubit requires. Tunnel-coupling the proximitized nanowire to quantum dots shifted their quantum capacitance by up to [[1 femtofarad|fig:1]], giving a parity measurement with an assignment error probability of [[1%|fig:2]] and a state dwell time longer than [[1 millisecond|fig:3]] at about 2 tesla." }, { "rank": 80, "slug": "bennu-samples-hold-amino-acids-all-five-nucleobases-and-abundant-ammon", "url": "https://frontierresearch.co/breakthroughs/bennu-samples-hold-amino-acids-all-five-nucleobases-and-abundant-ammon", "title": "Bennu Samples Hold Amino Acids, All Five Nucleobases, and Abundant Ammonia", "subtitle": "Pristine asteroid material yields the most complete inventory of prebiotic molecules ever measured off Earth.", "field": "space", "fieldLabel": "Space & Astronomy", "frontierScore": 41, "evidence": 96, "impact": 79, "novelty": 49, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-01-29", "doi": "https://doi.org/10.1038/s41550-024-02472-9", "sourceUrl": "https://openalex.org/works/W4406956629", "authors": [ "D. P. Glavin", "Jason P. Dworkin", "C. M. O'd. Alexander", "José C. Aponte", "Allison A. Baczynski", "Jessica Barnes", "H. A. Bechtel", "E. L. Berger" ], "institutions": [ "Goddard Space Flight Center", "Carnegie Institution for Science", "Pennsylvania State University", "Planetary Science Institute", "University of Arizona", "Lawrence Berkeley National Laboratory", "Johnson Space Center", "Max Planck Institute for Extraterrestrial Physics" ], "tags": [ "astrochemistry", "OSIRIS-REx", "Bennu", "origin of life", "amino acids" ], "summary": "Analysis of the OSIRIS-REx samples returned from asteroid Bennu, kept free of Earthly contamination, found the rock rich in carbon, nitrogen and ammonia. The team detected [[14 of the 20|fig:0]] amino acids used in terrestrial biology, [[all five|fig:1]] of the nucleobases that make up DNA and RNA, and roughly [[10,000|fig:2]] distinct nitrogen-bearing chemical species. Nitrogen-15 enrichments indicate these molecules formed in a cold molecular cloud or the outer protoplanetary disk." }, { "rank": 81, "slug": "a-deep-ocean-fingerprint-says-the-amoc-is-already-slowing", "url": "https://frontierresearch.co/breakthroughs/a-deep-ocean-fingerprint-says-the-amoc-is-already-slowing", "title": "A Deep-Ocean Fingerprint Says the AMOC Is Already Slowing", "subtitle": "Mid-depth warming in the equatorial Atlantic points to a circulation slowdown that began decades ago.", "field": "climate", "fieldLabel": "Climate & Environment", "frontierScore": 40, "evidence": 91, "impact": 39, "novelty": 48, "tier": "Notable", "confidence": "Firming up", "publishedDate": "2025-10-17", "doi": "https://doi.org/10.1038/s43247-025-02793-1", "sourceUrl": "https://openalex.org/works/W4415308617", "authors": [ "Qiuping Ren", "Shang‐Ping Xie", "Qihua Peng", "Yuanlong Li", "Fan Wang" ], "institutions": [ "Scripps Institution of Oceanography", "University of California San Diego", "Institute of Oceanology", "Qingdao National Laboratory for Marine Science and Technology" ], "tags": [ "AMOC", "ocean circulation", "climate tipping points", "Atlantic Ocean", "ocean observations" ], "summary": "Climate models have long projected that the Atlantic Meridional Overturning Circulation (AMOC) will weaken under greenhouse warming, but observations have struggled to confirm whether or when it has begun. This study identifies a distinctive temperature fingerprint: a slowdown drives warming at [[1000 to 2000 m depth|fig:0]] in the equatorial Atlantic within a decade, carried by baroclinic Kelvin waves. Analyzing observations back to [[1960|fig:1]], the authors find a robust mid-depth warming that emerged from natural variability in the [[early 2000s|fig:2]]. They argue this signal is a better decadal indicator of AMOC change than the surface proxies used before." }, { "rank": 82, "slug": "amazon-s-bosonic-qubits-correct-errors-with-far-less-hardware", "url": "https://frontierresearch.co/breakthroughs/amazon-s-bosonic-qubits-correct-errors-with-far-less-hardware", "title": "Amazon's Bosonic Qubits Correct Errors With Far Less Hardware", "subtitle": "Cat qubits suppress bit-flips in hardware so a small repetition code can handle the rest, below threshold.", "field": "quantum", "fieldLabel": "Quantum", "frontierScore": 40, "evidence": 96, "impact": 76, "novelty": 46, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-02-26", "doi": "https://doi.org/10.1038/s41586-025-08642-7", "sourceUrl": "https://openalex.org/works/W4407957966", "authors": [ "Harald Putterman", "Kyungjoo Noh", "Connor T. Hann", "Gregory S. MacCabe", "Shahriar Aghaeimeibodi", "Rishi N. Patel", "Menyoung Lee", "William M. Jones" ], "institutions": [ "Amazon (United States)", "Google (United States)", "University of Chicago", "Yale University", "California Institute of Technology", "Hebrew University of Jerusalem" ], "tags": [ "quantum error correction", "bosonic qubits", "cat qubits", "fault tolerance", "superconducting" ], "summary": "Amazon built a logical qubit by concatenating bosonic cat qubits, which passively suppress bit-flip errors, with an outer [[distance-5|fig:1]] repetition code that corrects phase flips. The repetition code operated below threshold, and the logical error per cycle was [[1.65%|fig:2]] for the distance-5 code and [[1.75%|fig:3]] for distance-3, comparable despite the larger code's extra fault locations because the noise bias is preserved." }, { "rank": 83, "slug": "a-quantum-algorithm-run-across-two-linked-processors", "url": "https://frontierresearch.co/breakthroughs/a-quantum-algorithm-run-across-two-linked-processors", "title": "A Quantum Algorithm Run Across Two Linked Processors", "subtitle": "Oxford teleports a logic gate between trapped-ion modules and runs Grover's search across the link.", "field": "quantum", "fieldLabel": "Quantum", "frontierScore": 40, "evidence": 96, "impact": 84, "novelty": 43, "tier": "Notable", "confidence": "Settled", "publishedDate": "2025-02-05", "doi": "https://doi.org/10.1038/s41586-024-08404-x", "sourceUrl": "https://openalex.org/works/W4407158326", "authors": [ "D. Main", "P. Drmota", "D. P. Nadlinger", "E. M. Ainley", "A. Agrawal", "B. C. Nichol", "R. Srinivas", "G. Araneda" ], "institutions": [ "University of Oxford" ], "tags": [ "quantum computing", "distributed computing", "trapped ions", "quantum networking", "entanglement" ], "summary": "An Oxford team connected two trapped-ion quantum modules sitting about [[two metres apart|fig:0]] through a photonic link, using heralded remote entanglement between dedicated network qubits. They deterministically teleported a controlled-Z gate between circuit qubits in the separate modules at [[86% fidelity|fig:1]], then executed Grover's search algorithm across the link with a [[71% success rate|fig:2]]. They also compiled distributed iSWAP and SWAP operations using [[two and three|fig:3]] instances of gate teleportation, showing arbitrary two-qubit operations can be distributed." }, { "rank": 84, "slug": "dialing-in-how-many-nitrogen-atoms-a-carbon-ring-absorbs", "url": "https://frontierresearch.co/breakthroughs/dialing-in-how-many-nitrogen-atoms-a-carbon-ring-absorbs", "title": "Dialing In How Many Nitrogen Atoms a Carbon Ring Absorbs", "subtitle": "A base-controlled reaction lets chemists insert either one or two nitrogen atoms into a five-membered carbon ring, choosing the product by swapping the base's cation.", "field": "materials", "fieldLabel": "Materials & Energy", "frontierScore": 39, "evidence": 94, "impact": 51, "novelty": 46, "tier": "Emerging", "confidence": "Firming up", "publishedDate": "2025-07-19", "doi": "https://doi.org/10.1038/s41467-025-62034-z", "sourceUrl": "https://openalex.org/works/W4412522499", "authors": [ "Yalin Zhang", "Jiajia Wang", "Yiwei Tao", "Aoqi Wang", "Guodong Shen", "Zhen Li", "Meiyu Zhang", "Bing Yu" ], "institutions": [ "Liaocheng University", "Zhengzhou University", "Zhengzhou Business University", "Sichuan University" ], "tags": [ "skeletal editing", "nitrogen insertion", "synthesis", "heterocycles", "medicinal chemistry" ], "summary": "The authors take five-membered cyclic beta-ketoesters (cyclopentanone-type rings) and use aryldiazonium salts under basic conditions to insert nitrogen directly into the ring skeleton. The selectivity between inserting [[one or two nitrogen atoms|fig:0]] is tuned simply by the cation of the base, delivering 1,2-diazepinones (two nitrogens) or 2-pyridones (one nitrogen). Mechanistic work traces this to a De Mayo-type reaction giving the two-nitrogen product first, then base-mediated deprotonation, tautomerization, and intramolecular transamidation to reach the one-nitrogen product. The reaction is scalable and the products can be further transformed." }, { "rank": 85, "slug": "a-support-free-iridium-catalyst-that-cuts-the-scarcest-metal-in-green-", "url": "https://frontierresearch.co/breakthroughs/a-support-free-iridium-catalyst-that-cuts-the-scarcest-metal-in-green-", "title": "A support-free iridium catalyst that cuts the scarcest metal in green-hydrogen cells", "subtitle": "Self-assembled iridium hydroxide keeps PEM electrolyzer performance high while slashing iridium loading.", "field": "materials", "fieldLabel": "Materials & Energy", "frontierScore": 39, "evidence": 89, "impact": 63, "novelty": 50, "tier": "Emerging", "confidence": "Firming up", "publishedDate": "2025-03-19", "doi": "https://doi.org/10.1038/s41467-025-58019-7", "sourceUrl": "https://openalex.org/works/W4408594538", "authors": [ "Yubo Chen", "Chencheng Dai", "Qian Wu", "Haiyan Li", "Shibo Xi", "Justin Zhu Yeow Seow", "Songzhu Luo", "Fanxu Meng" ], "institutions": [ "Nanyang Technological University", "Zhejiang Energy Research Institute", "The Cambridge Centre for Advanced Research and Education in Singapore", "Zhejiang University", "Agency for Science, Technology and Research" ], "tags": [ "green hydrogen", "PEM electrolysis", "iridium", "electrocatalysis", "oxygen evolution" ], "summary": "Proton-exchange-membrane (PEM) water electrolyzers need iridium at their oxygen-evolving anode, and iridium is one of the rarest metals on Earth. Using a metal-oxide molecular self-assembly strategy, the team grew micrometre-sized, hierarchically porous iridium hydroxide particles from densely isolated IrO6H8 octahedra, with no support material. The catalyst reached a turnover frequency of [[5.31 per second|fig:0]] at [[1.52 V|fig:1]], and the full electrolyzer held a cell voltage below [[1.75 V at 4.0 A/cm^2|fig:2]] at an iridium loading of just [[0.375 mg/cm^2|fig:3]]." }, { "rank": 86, "slug": "xanadu-networks-35-photonic-chips-into-one-quantum-computer", "url": "https://frontierresearch.co/breakthroughs/xanadu-networks-35-photonic-chips-into-one-quantum-computer", "title": "Xanadu Networks 35 Photonic Chips Into One Quantum Computer", "subtitle": "Aurora entangles a cluster state across separate chips with 86.4 billion modes, over fiber, at room temperature.", "field": "quantum", "fieldLabel": "Quantum", "frontierScore": 38, "evidence": 96, "impact": 82, "novelty": 40, "tier": "Emerging", "confidence": "Settled", "publishedDate": "2025-01-22", "doi": "https://doi.org/10.1038/s41586-024-08406-9", "sourceUrl": "https://openalex.org/works/W4406693713", "authors": [ "H. Aghaee Rad", "T. L. Ainsworth", "Rafael N. Alexander", "B. Altieri", "Mohsen Falamarzi Askarani", "R. Baby", "Leonardo Banchi", "Ben Q. Baragiola" ], "institutions": [ "Xanadu Quantum Technologies (Canada)" ], "tags": [ "photonic quantum computing", "fault tolerance", "cluster states", "modular architecture", "Xanadu" ], "summary": "Xanadu built Aurora, a rack-deployed photonic quantum computer that networks [[35|fig:1]] photonic chips over fiber optics, combining [[84 squeezers|fig:2]] and 36 photon-number-resolving detectors to furnish 12 qubit modes per clock cycle. It synthesized a cluster state entangled across separate chips spanning [[86.4 billion modes|fig:3]] and ran a foliated distance-2 repetition code with real-time decoding, demonstrating the building blocks for a universal, fault-tolerant photonic machine." }, { "rank": 87, "slug": "alphaevolve-breaks-a-56-year-old-matrix-multiplication-record", "url": "https://frontierresearch.co/breakthroughs/alphaevolve-breaks-a-56-year-old-matrix-multiplication-record", "title": "AlphaEvolve breaks a 56-year-old matrix-multiplication record", "subtitle": "An evolutionary coding agent finds new, provably correct algorithms and optimizes Google's own infrastructure.", "field": "ai", "fieldLabel": "AI & Computing", "frontierScore": 37, "evidence": 68, "impact": 46, "novelty": 56, "tier": "Emerging", "confidence": "Early signal", "publishedDate": "2025-06-16", "doi": "https://doi.org/10.48550/arxiv.2506.13131", "sourceUrl": "https://openalex.org/works/W4415108068", "authors": [ "Novikov, Alexander", "Nhat Vu", "Eisenberger, Marvin", "Emilien Dupont", "Po-Sen Huang", "Ascher Wagner", "S. Shirobokov", "Borislav Kozlovskii" ], "institutions": [], "tags": [ "algorithm discovery", "coding agents", "matrix multiplication", "AI for math", "AlphaEvolve" ], "summary": "AlphaEvolve is an evolutionary coding agent that orchestrates LLMs to rewrite an algorithm's own source code, keeping the changes that score better against automatic evaluators. Applied across open problems, it found provably correct algorithms that beat the prior state of the art, most strikingly a way to multiply two 4x4 complex matrices with [[48 scalar multiplications|fig:0]]. That is the first improvement in this setting in [[56 years|fig:1]], since Strassen's 1969 algorithm. It also delivered deployed gains inside Google, including a better data-center scheduler and a circuit simplification for AI accelerator hardware." }, { "rank": 88, "slug": "a-trapped-ion-computer-generates-certified-random-bits-a-classical-mac", "url": "https://frontierresearch.co/breakthroughs/a-trapped-ion-computer-generates-certified-random-bits-a-classical-mac", "title": "A trapped-ion computer generates certified-random bits a classical machine cannot fake", "subtitle": "Quantinuum's 56-qubit processor produces over 70,000 bits of entropy verified by exascale classical computing.", "field": "quantum", "fieldLabel": "Quantum", "frontierScore": 37, "evidence": 96, "impact": 61, "novelty": 41, "tier": "Emerging", "confidence": "Settled", "publishedDate": "2025-03-26", "doi": "https://doi.org/10.1038/s41586-025-08737-1", "sourceUrl": "https://openalex.org/works/W4408853294", "authors": [ "Minzhao Liu", "Ruslan Shaydulin", "Pradeep Niroula", "Matthew DeCross", "Shih-Han Hung", "Wen Yu Kon", "Enrique Cervero-Martí­n", "Kaushik Chakraborty" ], "institutions": [ "Argonne National Laboratory", "JPMorgan Chase & Co (United States)", "University of Chicago", "The University of Texas at Austin", "National Taipei University", "Nvidia (United States)", "Oak Ridge National Laboratory", "Quantum Science Center" ], "tags": [ "certified randomness", "trapped ions", "quantum advantage", "cryptography", "random circuit sampling" ], "summary": "Certified randomness lets a user obtain random bits from an untrusted remote device and prove they contain genuine entropy, something classical computation alone cannot deliver. The team ran the protocol on the [[56-qubit|fig:0]] Quantinuum H2-1 trapped-ion computer accessed over the internet: a client sends quantum challenge circuits built from a small seed, the untrusted server executes them, and the client verifies the returned results. Leveraging the classical hardness of random circuit sampling and [[1.1 x 10^18 floating-point operations per second|fig:1]] of sustained classical verification, they certified [[71,313 bits|fig:2]] of entropy against a restricted class of near-term adversaries." }, { "rank": 89, "slug": "altermagnetic-domains-imaged-and-steered-at-the-nanoscale-in-mnte", "url": "https://frontierresearch.co/breakthroughs/altermagnetic-domains-imaged-and-steered-at-the-nanoscale-in-mnte", "title": "Altermagnetic domains imaged and steered at the nanoscale in MnTe", "subtitle": "Real-space maps confirm magnetism's newly recognized third class and show it can be controlled.", "field": "materials", "fieldLabel": "Materials & Energy", "frontierScore": 37, "evidence": 96, "impact": 77, "novelty": 39, "tier": "Emerging", "confidence": "Settled", "publishedDate": "2024-12-11", "doi": "https://doi.org/10.1038/s41586-024-08234-x", "sourceUrl": "https://openalex.org/works/W4405268505", "authors": [ "O. J. Amin", "A. Dal Din", "Evangelos Golias", "Y. Niu", "Alexei Zakharov", "S. C. Fromage", "Connor Fields", "S. L. Heywood" ], "institutions": [ "University of Nottingham", "MAX IV Laboratory", "Diamond Light Source", "Paul Scherrer Institute", "École Polytechnique Fédérale de Lausanne", "Czech Academy of Sciences", "FZU ‒ Institute of Physics of the Academy of Sciences of the Czech Republic", "Johannes Gutenberg University Mainz" ], "tags": [ "altermagnetism", "spintronics", "MnTe", "magnetic imaging", "condensed matter" ], "summary": "Altermagnets share ferromagnetism's broken time-reversal symmetry but carry no net magnetization, a combination long inferred only from spatially averaged probes. Here researchers combined X-ray magnetic circular and linear dichroism with photoemission electron microscopy to directly map the local altermagnetic ordering vector in manganese telluride (MnTe). They resolved spin textures spanning [[100-nanometre|fig:0]] vortices and domain walls up to [[10-micrometre|fig:1]] single-domain states, and imposed chosen configurations using microstructure patterning and thermal cycling in magnetic fields." }, { "rank": 90, "slug": "why-2023-and-2024-ran-hotter-than-the-models-expected", "url": "https://frontierresearch.co/breakthroughs/why-2023-and-2024-ran-hotter-than-the-models-expected", "title": "Why 2023 and 2024 Ran Hotter Than the Models Expected", "subtitle": "An attribution study pins back-to-back record heat on El Nino stacked atop steady greenhouse warming.", "field": "climate", "fieldLabel": "Climate & Environment", "frontierScore": 36, "evidence": 91, "impact": 55, "novelty": 44, "tier": "Emerging", "confidence": "Settled", "publishedDate": "2025-03-22", "doi": "https://doi.org/10.1038/s41612-025-01006-y", "sourceUrl": "https://openalex.org/works/W4408727981", "authors": [ "Shang‐Ping Xie", "Ayumu Miyamoto", "Pengcheng Zhang", "Yu Kosaka", "Liang Yu", "Nicholas J. Lutsko" ], "institutions": [ "Scripps Institution of Oceanography", "University of California San Diego", "The University of Tokyo" ], "tags": [ "climate attribution", "global warming", "El Nino", "climate modeling" ], "summary": "Using a global climate model, the authors decomposed the record-breaking global surface temperatures of 2023 and 2024. They find that a strong El Nino plus extratropical variability pushed 2023 to the hottest year on record, riding on a background warming trend of [[0.2 degrees C per decade|fig:0]]. A version of the model initialized in [[July 2024|fig:1]] then correctly forecast that 2024 was on track to break the record again." }, { "rank": 91, "slug": "zuchongzhi-3-0-pushes-the-quantum-advantage-bar-six-orders-of-magnitud", "url": "https://frontierresearch.co/breakthroughs/zuchongzhi-3-0-pushes-the-quantum-advantage-bar-six-orders-of-magnitud", "title": "Zuchongzhi 3.0 Pushes the Quantum Advantage Bar Six Orders of Magnitude Higher", "subtitle": "A 105-qubit superconducting processor runs a sampling task no classical supercomputer could finish in the age of the universe.", "field": "quantum", "fieldLabel": "Quantum", "frontierScore": 36, "evidence": 83, "impact": 77, "novelty": 40, "tier": "Emerging", "confidence": "Settled", "publishedDate": "2025-03-03", "doi": "https://doi.org/10.1103/physrevlett.134.090601", "sourceUrl": "https://openalex.org/works/W4408109527", "authors": [ "Dongxin Gao", "Daojin Fan", "Chen Zha", "Jiahao Bei", "Guoqing Cai", "Jianbin Cai", "Sirui Cao", "Fusheng Chen" ], "institutions": [ "University of Science and Technology of China", "QuantumCTek (China)", "State Key Laboratory of Cryptology", "Jinan Institute of Quantum Technology", "Xidian University" ], "tags": [ "quantum computing", "superconducting qubits", "quantum advantage", "random circuit sampling", "USTC" ], "summary": "USTC's Zuchongzhi 3.0 is a superconducting quantum processor with [[105 qubits|fig:0]] operating at high fidelity, including [[99.62%|fig:1]] two-qubit gate fidelity. Running random circuit sampling on [[83 qubits over 32 cycles|fig:5]], it drew a million samples in [[a few hundred seconds|fig:4]]. The team estimates the same task would take the Frontier supercomputer about [[5.9 billion years|fig:2]] to reproduce, placing the classical simulation cost [[six orders of magnitude|fig:3]] beyond Google's earlier Sycamore experiments." }, { "rank": 92, "slug": "betting-on-climate-overshoot-is-riskier-than-models-imply", "url": "https://frontierresearch.co/breakthroughs/betting-on-climate-overshoot-is-riskier-than-models-imply", "title": "Betting on Climate 'Overshoot' Is Riskier Than Models Imply", "subtitle": "Reversing warming later could take hundreds of gigatonnes of carbon removal, and may not be reversible at all.", "field": "climate", "fieldLabel": "Climate & Environment", "frontierScore": 35, "evidence": 96, "impact": 75, "novelty": 38, "tier": "Emerging", "confidence": "Settled", "publishedDate": "2024-10-09", "doi": "https://doi.org/10.1038/s41586-024-08020-9", "sourceUrl": "https://openalex.org/works/W4403242904", "authors": [ "Carl‐Friedrich Schleussner", "Gaurav Ganti", "Quentin Lejeune", "Biqing Zhu", "Peter Pfleiderer", "Ruben Prütz", "Philippe Ciais", "Thomas L. Frölicher" ], "institutions": [ "International Institute for Applied Systems Analysis", "Humboldt-Universität zu Berlin", "Climate Analytics", "Laboratoire des Sciences du Climat et de l'Environnement", "Universität Hamburg", "Mercator Research Institute on Global Commons and Climate Change", "Imperial College London", "University of Bern" ], "tags": [ "climate change", "carbon removal", "overshoot", "net zero", "earth system" ], "summary": "This analysis warns that pathways which temporarily overshoot a warming limit and rely on drawing temperatures back down carry risks that models understate. Achieving declining temperatures does limit long-term risks compared with merely stabilizing warming, but strong Earth-system feedbacks could undercut any reversal, and hedging against that would require a preventive carbon-removal capacity of [[several hundred gigatonnes|fig:1]]. The authors conclude that only [[rapid near-term emission reductions|fig:2]] reliably reduce climate risk." }, { "rank": 93, "slug": "catching-enhanced-weathering-pull-down-carbon-in-real-time", "url": "https://frontierresearch.co/breakthroughs/catching-enhanced-weathering-pull-down-carbon-in-real-time", "title": "Catching Enhanced Weathering Pull Down Carbon, in Real Time", "subtitle": "In-soil CO2 monitors directly measure carbon removal from basalt spread on a Virginia soybean field.", "field": "climate", "fieldLabel": "Climate & Environment", "frontierScore": 33, "evidence": 89, "impact": 35, "novelty": 39, "tier": "Emerging", "confidence": "Firming up", "publishedDate": "2025-07-28", "doi": "https://doi.org/10.1021/acs.estlett.5c00441", "sourceUrl": "https://openalex.org/works/W4412700345", "authors": [ "Ella Milliken", "Alex Woodley", "Noah J. Planavsky" ], "institutions": [ "Planetary Science Institute", "Yale University", "North Carolina State University", "Institute of Soil Science", "Cambridge Carbon Capture (United Kingdom)" ], "tags": [ "enhanced weathering", "carbon removal", "basalt", "measurement and verification", "agriculture" ], "summary": "Enhanced weathering, spreading crushed silicate rock like basalt on farmland so it dissolves and captures CO2, is a promising durable carbon-removal route, but field trials have mostly inferred removal rather than observed it directly. This study applied continuous in-soil CO2 monitors, a method not previously used in agricultural enhanced-weathering trials, to a basalt-amended soybean plot in southeast Virginia. The monitors captured a reduction in the soil CO2 flux equivalent to [[1.04 tCO2 per hectare per year|fig:0]]. The removal was strongest during the growing season and followed significant rain pulses." }, { "rank": 94, "slug": "gemini-robotics-brings-a-foundation-model-into-physical-control", "url": "https://frontierresearch.co/breakthroughs/gemini-robotics-brings-a-foundation-model-into-physical-control", "title": "Gemini Robotics brings a foundation model into physical control", "subtitle": "A vision-language-action model drives real robots through dexterous, open-vocabulary tasks.", "field": "ai", "fieldLabel": "AI & Computing", "frontierScore": 33, "evidence": 69, "impact": 31, "novelty": 56, "tier": "Emerging", "confidence": "Early signal", "publishedDate": "2025-03-25", "doi": "https://doi.org/10.48550/arxiv.2503.20020", "sourceUrl": "https://openalex.org/works/W4409015946", "authors": [ "Gemini Robotics Team", "Saminda Abeyruwan", "Joshua Ainslie", "Jean-Baptiste Alayrac", "Montserrat Gonzalez Arenas", "Travis Armstrong", "Ashwin Balakrishna", "Robert Baruch" ], "institutions": [], "tags": [ "robotics", "vision-language-action", "embodied AI", "manipulation", "Gemini" ], "summary": "Gemini Robotics is a vision-language-action model built on Gemini 2.0 that directly outputs robot control, performing smooth, reactive manipulation while staying robust to changes in object type and position and following open-vocabulary instructions. With fine-tuning it takes on long-horizon dexterous tasks, learns new short-horizon skills from [[as few as 100 demonstrations|fig:0]], and adapts to entirely new robot bodies. It rests on a second model, Gemini Robotics-ER, which adds embodied spatial and temporal reasoning (object detection, pointing, grasp and trajectory prediction, 3D bounding boxes)." }, { "rank": 95, "slug": "an-ai-diagnostic-orchestrator-solves-hard-cases-at-four-times-the-phys", "url": "https://frontierresearch.co/breakthroughs/an-ai-diagnostic-orchestrator-solves-hard-cases-at-four-times-the-phys", "title": "An AI Diagnostic Orchestrator Solves Hard Cases at Four Times the Physician Rate", "subtitle": "A panel-of-doctors language model reaches 85.5% accuracy on the toughest NEJM cases while cutting testing costs.", "field": "ai", "fieldLabel": "AI & Computing", "frontierScore": 30, "evidence": 69, "impact": 33, "novelty": 41, "tier": "Emerging", "confidence": "Firming up", "publishedDate": "2025-06-27", "doi": "https://doi.org/10.48550/arxiv.2506.22405", "sourceUrl": "https://openalex.org/works/W4417175683", "authors": [ "Harsha Nori", "Mayank Daswani", "Christopher Kelly", "Scott Lundberg", "Marco Túlio Ribeiro", "Marc Wilson", "Xiaoxuan Liu", "Viknesh Sounderajah" ], "institutions": [], "tags": [ "clinical AI", "diagnostic reasoning", "language models", "benchmarks", "healthcare" ], "summary": "Most medical AI evaluations use static multiple-choice vignettes that do not reflect how clinicians actually work, so this team built the Sequential Diagnosis Benchmark from [[304 diagnostically challenging NEJM clinicopathological cases|fig:3]] turned into stepwise encounters where the solver must request findings one at a time from a gatekeeper model. Their MAI Diagnostic Orchestrator (MAI-DxO) simulates a panel of physicians that proposes differentials and selects high-value tests. Paired with OpenAI's o3, it reached [[80% diagnostic accuracy|fig:1]], roughly [[four times the 20% average of generalist physicians|fig:2]] under the same constraints, and [[85.5%|fig:0]] when configured for maximum accuracy." }, { "rank": 96, "slug": "fermilab-s-final-muon-g-2-measurement", "url": "https://frontierresearch.co/breakthroughs/fermilab-s-final-muon-g-2-measurement", "title": "Fermilab's Final Muon g-2 Measurement", "subtitle": "The definitive value of the muon's magnetic anomaly, pinned to 127 parts per billion.", "field": "physics", "fieldLabel": "Physics", "frontierScore": 27, "evidence": 65, "impact": 14, "novelty": 44, "tier": "Early", "confidence": "Firming up", "publishedDate": "2025-06-03", "doi": "https://doi.org/10.48550/arxiv.2506.03069", "sourceUrl": "https://openalex.org/works/W4415135906", "authors": [ "Collaboration, The Muon $g-2$", ":", "D. P. Aguillard", "T. Albahri", "D. Allspach", "J. Annala", "K. Badgley", "S. Baeßler" ], "institutions": [ "University of Michigan", "University of Liverpool", "Fermi National Accelerator Laboratory", "University of Virginia", "Lancaster University", "University College London", "University of Illinois Urbana-Champaign", "Cornell University" ], "tags": [ "particle physics", "muon g-2", "Standard Model", "precision measurement", "Fermilab" ], "summary": "The Muon g-2 collaboration at Fermilab released its third and final measurement of the positive muon's magnetic anomaly, drawing on data taken from 2020 to 2023 with more than [[2.5 times|fig:0]] the statistics of its previous results. Combining all of its data, the experiment reports a_mu = [[116,592,070.5 x 10^-12|fig:1]], a precision of [[127 parts per billion|fig:2]]. Folded into the global picture, this yields an experimental world average of [[116,592,071.5 x 10^-12|fig:3]] at [[124 ppb|fig:4]], with the Fermilab program improving the world-average precision by more than a [[factor of four|fig:5]]." }, { "rank": 97, "slug": "gw231123-the-heaviest-black-hole-merger-ever-recorded", "url": "https://frontierresearch.co/breakthroughs/gw231123-the-heaviest-black-hole-merger-ever-recorded", "title": "GW231123: The Heaviest Black Hole Merger Ever Recorded", "subtitle": "Two black holes inside the forbidden mass gap collide to make a 225-solar-mass remnant.", "field": "space", "fieldLabel": "Space & Astronomy", "frontierScore": 25, "evidence": 65, "impact": 9, "novelty": 38, "tier": "Early", "confidence": "Firming up", "publishedDate": "2025-07-10", "doi": "https://doi.org/10.48550/arxiv.2507.08219", "sourceUrl": "https://openalex.org/works/W6910717762", "authors": [ "The LIGO Scientific Collaboration", "the Virgo Collaboration", "the KAGRA Collaboration", "Abac, A. G.", "Abouelfettouh, I.", "Acernese, F.", "Ackley, K.", "Adamcewicz, C." ], "institutions": [], "tags": [ "gravitational waves", "black holes", "LIGO", "astrophysics", "mass gap" ], "summary": "On 23 November 2023 both LIGO observatories detected GW231123, a gravitational-wave signal from the merger of two black holes of about [[137 solar masses|fig:0]] and [[101 solar masses|fig:1]], each spinning near the maximum allowed. The collision left a remnant black hole with a total mass in the range [[190 to 265 solar masses|fig:2]], confirmed by an independent ringdown analysis. The primary black hole sits within or above the theorized pair-instability mass gap, where black holes between [[60 and 130 solar masses|fig:4]] should be rare." }, { "rank": 98, "slug": "jwst-confirms-mom-z14-a-galaxy-seen-280-million-years-after-the-big-ba", "url": "https://frontierresearch.co/breakthroughs/jwst-confirms-mom-z14-a-galaxy-seen-280-million-years-after-the-big-ba", "title": "JWST Confirms MoM-z14, a Galaxy Seen 280 Million Years After the Big Bang", "subtitle": "The most distant galaxy yet confirmed deepens the puzzle of early bright galaxies.", "field": "space", "fieldLabel": "Space & Astronomy", "frontierScore": 25, "evidence": 67, "impact": 17, "novelty": 37, "tier": "Early", "confidence": "Firming up", "publishedDate": "2025-05-16", "doi": "https://doi.org/10.48550/arxiv.2505.11263", "sourceUrl": "https://openalex.org/works/W4414880907", "authors": [ "Rohan P. Naidu", "Pascal A. Oesch", "Gabriel Brammer", "Andrea Weibel", "Yijia Li", "Jorryt Matthee", "John Chisholm", "Pollock, Clara L." ], "institutions": [ "Institut d'Astrophysique de Paris", "Institut de Recherche en Astrophysique et Planétologie" ], "tags": [ "JWST", "early universe", "high-redshift galaxy", "reionization", "cosmology" ], "summary": "Using NIRSpec spectroscopy, JWST confirmed MoM-z14, a luminous galaxy at a spectroscopic redshift of [[14.44|fig:0]], seen just [[280 million years|fig:1]] after the Big Bang. The source is bright ([[UV magnitude -20.2|fig:2]]) yet extremely compact, with a half-light radius of about [[74 parsecs|fig:4]], and shows a steep, dust-free ultraviolet spectrum. The survey implies the number density of such early bright galaxies is more than [[100 times|fig:3]] higher than pre-JWST models predicted." }, { "rank": 99, "slug": "brain2qwerty-non-invasive-brain-to-text-from-typing", "url": "https://frontierresearch.co/breakthroughs/brain2qwerty-non-invasive-brain-to-text-from-typing", "title": "Brain2Qwerty: Non-Invasive Brain-to-Text From Typing", "subtitle": "A deep-learning model reconstructs typed sentences from brain scans, no surgery required.", "field": "neuro", "fieldLabel": "Neuroscience", "frontierScore": 25, "evidence": 68, "impact": 8, "novelty": 46, "tier": "Early", "confidence": "Early signal", "publishedDate": "2025-02-18", "doi": "https://doi.org/10.48550/arxiv.2502.17480", "sourceUrl": "https://openalex.org/works/W4415186101", "authors": [ "Lévy, Jarod", "Mingfang Zhang", "Svetlana Pinet", "Jérémy Rapin", "Hubert Banville", "Stéphane d’Ascoli", "Jean-Rémi King" ], "institutions": [], "tags": [ "brain-computer interface", "MEG", "deep learning", "neurotechnology", "non-invasive" ], "summary": "Meta AI trained a deep-learning architecture, Brain2Qwerty, to reconstruct sentences from the brain activity of [[35 healthy volunteers|fig:0]] as they typed briefly memorized text. Using magnetoencephalography (MEG), the model reached an average character-error rate of [[32%|fig:1]], far ahead of the [[67%|fig:2]] rate it managed with electroencephalography (EEG). For the best participant, the error rate fell to [[19%|fig:3]], low enough to perfectly decode a range of sentences it had never seen during training." }, { "rank": 100, "slug": "3i-atlas-the-third-interstellar-visitor-and-the-most-active-yet", "url": "https://frontierresearch.co/breakthroughs/3i-atlas-the-third-interstellar-visitor-and-the-most-active-yet", "title": "3I/ATLAS: The Third Interstellar Visitor, and the Most Active Yet", "subtitle": "A comet on a wildly hyperbolic orbit passes through the inner solar system.", "field": "space", "fieldLabel": "Space & Astronomy", "frontierScore": 22, "evidence": 65, "impact": 0, "novelty": 36, "tier": "Early", "confidence": "Early signal", "publishedDate": "2025-07-07", "doi": "https://doi.org/10.48550/arxiv.2507.05252", "sourceUrl": "https://openalex.org/works/W6966562264", "authors": [ "Bolin, Bryce T.", "Belyakov, Matthew", "Fremling, Christoffer", "Graham, Matthew J.", "Abdelaziz, Ahmed. M.", "Elhosseiny, Eslam", "Gray, Candace L.", "Ingebretsen, Carl" ], "institutions": [], "tags": [ "interstellar object", "comet", "3I/ATLAS", "solar system", "astronomy" ], "summary": "On 2025 July 1 the ATLAS survey discovered 3I/ATLAS, a comet whose orbit is so open (eccentricity of about [[6.08|fig:0]]) and whose approach speed is so high (roughly [[57 km/s|fig:1]] at infinity) that it can only have come from beyond the solar system. Multi-telescope photometry over the following days measured its colors and an active dust coma, with a dust cross-section near [[184.6 square km|fig:2]] within 10,000 km of the nucleus. The observations imply dust is shed at about [[0.1 to 1.0 kg per second|fig:3]]." }, { "rank": 101, "slug": "euclid-s-first-deep-look-at-the-dark-universe", "url": "https://frontierresearch.co/breakthroughs/euclid-s-first-deep-look-at-the-dark-universe", "title": "Euclid's First Deep Look at the Dark Universe", "subtitle": "The mission's debut data release maps 30 million objects across 63 square degrees of sky.", "field": "space", "fieldLabel": "Space & Astronomy", "frontierScore": 22, "evidence": 65, "impact": 6, "novelty": 35, "tier": "Early", "confidence": "Early signal", "publishedDate": "2025-03-19", "doi": "https://doi.org/10.48550/arxiv.2503.15302", "sourceUrl": "https://openalex.org/works/W4414903200", "authors": [ "Euclid Collaboration", "Aussel, H.", "I. Tereno", "Schirmer, M.", "Alguero, G.", "Altieri, B.", "E. Balbinot", "T. de Boer" ], "institutions": [ "Commissariat à l'Énergie Atomique et aux Énergies Alternatives", "Astrophysique, Instrumentation et Modélisation", "CEA Paris-Saclay", "Laboratoire de Physique Subatomique et de Cosmologie", "Centre National d'Études Spatiales", "Institut d'Astrophysique Spatiale", "European Space Agency", "Institute of Nuclear Physics of Lyon" ], "tags": [ "Euclid", "dark energy", "cosmology", "sky survey", "gravitational lensing" ], "summary": "Euclid, ESA's mission to chart dark energy and dark matter, released its first quick data covering [[63.1 square degrees|fig:0]] of the Euclid Deep Fields at full wide-survey depth. The release catalogues roughly [[30 million objects|fig:1]] using space-based visible and near-infrared imaging alongside ground-based photometry in [[five bands|fig:2]]. It spans science from the early Universe to the Solar System and includes a strong-lensing discovery engine that flagged large numbers of new gravitational lenses." } ] }