{"days":[
{"date":"2026-05-23","headline":"First edition published; nearly all chapters reach full depth the same day","entries":[{"category":"milestone","title":"First edition of The Life Sciences AI Handbook published","change":"Publishes the initial handbook with chapters organized into foundations, molecular AI, cellular systems, therapeutics AI, engineering and automation, and practice and governance, plus appendices and front matter.","why":"Opens a dedicated evidence-based reference on AI for biomedical discovery, biotechnology, and translational research, alongside the companion handbooks.","chapters":[["Welcome","/"],["Biological Data Infrastructure","/foundations/data-infrastructure.html"],["Antibody and Biologic Design","/molecular/antibodies.html"],["Microbiome and Multi-Omics AI","/cells/multiomics.html"],["mRNA, RNA, and Vaccine Design","/therapeutics/vaccines.html"],["Agentic Science Workflows","/automation/agentic-workflows.html"],["Reproducibility and Open Science","/governance/reproducibility.html"],["Glossary","/appendices/glossary.html"]],"sources":[{"label":"AlphaFold Protein Structure Database, 2026","url":"https://alphafold.ebi.ac.uk/about","type":"other"},{"label":"Watson et al., 2023","url":"https://www.nature.com/articles/s41586-023-06415-8","type":"paper"},{"label":"Dauparas et al., 2022","url":"https://doi.org/10.1126/science.add2187","type":"paper"}]},{"category":"revision","title":"Foundations and Protein Structure Prediction: full-depth expansion","change":"Foundations, Protein Structure Prediction, and the Welcome page draw on dozens of primary sources spanning structure prediction, autonomous experimentation, and the foundation-model field.","why":"Brings the handbook's launch chapters to the same evidence depth as the companion handbooks.","chapters":[["AI for the Life Sciences","/foundations/ai-for-life-sciences.html"],["Protein Structure Prediction","/molecular/protein-structure.html"],["Welcome","/"]],"sources":[{"label":"Varadi et al., 2023","url":"https://doi.org/10.1093/nar/gkad1011","type":"paper"},{"label":"Boiko et al., 2023","url":"https://doi.org/10.1038/s41586-023-06792-0","type":"paper"},{"label":"Abramson et al., 2024","url":"https://doi.org/10.1038/s41586-024-07487-w","type":"paper"}]},{"category":"content","title":"History of AI in the Life Sciences","change":"Adds a chapter tracing computational biology's use of learning algorithms from the algorithmic era (1970 to 1990) and statistical learning era (1990 to 2012) through today's foundation models, and reorders the book's parts.","why":"Gives readers historical context for how the field reached its current foundation-model era, and reorders the book's parts for long-term durability.","chapters":[["History of AI in the Life Sciences","/foundations/history.html"]],"sources":[{"label":"Altschul et al., 1990","url":"https://doi.org/10.1016/S0022-2836(05)80360-2","type":"paper"},{"label":"Alipanahi et al., 2015","url":"https://doi.org/10.1038/nbt.3300","type":"paper"},{"label":"Poplin et al., 2018","url":"https://doi.org/10.1038/nbt.4235","type":"paper"}]},{"category":"revision","title":"Evaluation Principles and Foundation Models for Biology deepened","change":"Expands Evaluation Principles with sections on blind benchmarks and biology-aware data splits, and expands Foundation Models for Biology with sections tracing the protein language model lineage from ESM-1b through ESM3.","why":"Establishes the evaluation standards and the shared foundation-model pattern the rest of the handbook builds on.","chapters":[["Evaluation Principles for Life Sciences AI","/foundations/evaluation-principles.html"],["Foundation Models for Biology","/foundations/foundation-models.html"]],"sources":[{"label":"Wu et al., 2018","url":"https://doi.org/10.1039/C7SC02664A","type":"paper"},{"label":"Walsh et al., 2021","url":"https://doi.org/10.1038/s41592-021-01205-4","type":"paper"},{"label":"Rives et al., 2021","url":"https://doi.org/10.1073/pnas.2016239118","type":"paper"},{"label":"Elnaggar et al., 2022","url":"https://doi.org/10.1109/TPAMI.2021.3095381","type":"paper"}]},{"category":"revision","title":"Protein Design and Variant Effect Prediction deepened","change":"Expands Protein Design with sections on backbone generation (RFdiffusion) and sequence design (ProteinMPNN, LigandMPNN), and expands Variant Effect Prediction with missense, splicing, and non-coding variant sections, together adding 18 verified citations.","why":"Documents the generative design and variant-interpretation methods that complement structure prediction in the molecular AI chapters.","chapters":[["Protein Design and Engineering","/molecular/protein-design.html"],["Variant Effect Prediction","/molecular/variant-effects.html"]],"sources":[{"label":"Dauparas et al., 2025","url":"https://doi.org/10.1038/s41592-025-02626-1","type":"paper"},{"label":"Ingraham et al., 2023","url":"https://doi.org/10.1038/s41586-023-06728-8","type":"paper"},{"label":"Frazer et al., 2021","url":"https://doi.org/10.1038/s41586-021-04043-8","type":"paper"},{"label":"Sundaram et al., 2018","url":"https://doi.org/10.1038/s41588-018-0167-z","type":"paper"}]},{"category":"revision","title":"Single-Cell Foundation Models deepened","change":"Single-Cell Foundation Models covers scGPT, Geneformer, scFoundation, scBERT, UCE, and tGPT, with sections on representation learning at scale, cell-atlas search, and integration and batch correction, drawing on 17 new citations.","why":"Surveys the wave of single-cell foundation models trained between 2022 and 2024 so readers can compare their scale and validation.","chapters":[["Single-Cell Foundation Models","/cells/single-cell-models.html"]],"sources":[{"label":"Hao et al., 2024","url":"https://doi.org/10.1038/s41592-024-02305-7","type":"paper"},{"label":"Yang et al., 2022","url":"https://doi.org/10.1038/s42256-022-00534-z","type":"paper"},{"label":"Shen et al., 2023","url":"https://doi.org/10.1016/j.isci.2023.106536","type":"paper"}]},{"category":"revision","title":"Small Molecule Generation and ADMET deepened","change":"Expands the chapter with sections on generative chemistry, structure-based docking, and property prediction, tracing the REINVENT lineage from 2017 through 2024 among 18 new citations.","why":"Maps the stacked AI layers used in small-molecule drug discovery, from candidate generation through property prediction.","chapters":[["Small Molecule Generation and ADMET","/therapeutics/small-molecules.html"]],"sources":[{"label":"Olivecrona et al., 2017","url":"https://doi.org/10.1186/s13321-017-0235-x","type":"paper"},{"label":"Blaschke et al., 2020","url":"https://doi.org/10.1021/acs.jcim.0c00915","type":"paper"},{"label":"Loeffler et al., 2024","url":"https://doi.org/10.1186/s13321-024-00812-5","type":"paper"}]},{"category":"revision","title":"Self-Driving Laboratories deepened","change":"Expands the chapter with sections on chemistry automation, biology automation, and the discipline needed to evaluate novelty claims from systems such as A-Lab, alongside Coscientist and the Virtual Lab.","why":"Distinguishes the more mature chemistry-automation track from the earlier-stage biology track in closed-loop experimentation.","chapters":[["Self-Driving Laboratories","/automation/self-driving-labs.html"]],"sources":[{"label":"Szymanski et al., 2023","url":"https://doi.org/10.1038/s41586-023-06734-w","type":"paper"},{"label":"MacLeod et al., 2020","url":"https://doi.org/10.1126/sciadv.aaz8867","type":"paper"},{"label":"Roch et al., 2020","url":"https://doi.org/10.1371/journal.pone.0229862","type":"paper"}]},{"category":"content","title":"Toolkit and Emerging Frontiers chapters","change":"Adds a Practical Workflows part with a Toolkit chapter on selecting tools for structure prediction and sequence search, and a Future Directions part with an Emerging Frontiers chapter surveying developing capabilities such as virtual cells.","why":"Completes the handbook's eight-part structure with practical tool guidance and a forward-looking survey of developing capabilities.","chapters":[["Toolkit for AI-Augmented Bio Research","/practical/toolkit.html"],["Emerging Frontiers in AI for the Life Sciences","/future/emerging.html"]],"sources":[{"label":"Mirdita et al., 2022","url":"https://doi.org/10.1038/s41592-022-01488-1","type":"paper"},{"label":"Buchfink et al., 2014","url":"https://doi.org/10.1038/nmeth.3176","type":"paper"},{"label":"Bunne et al., 2024","url":"https://doi.org/10.1016/j.cell.2024.11.015","type":"paper"},{"label":"Wiens et al., 2019","url":"https://doi.org/10.1038/s41591-019-0548-6","type":"paper"}]},{"category":"revision","title":"Remaining chapters expanded across the book","change":"Expands the 18 remaining chapters across the cellular systems, therapeutics, automation, and governance areas, for example Cell Painting and Image-Based Phenotyping and Benchmarks, adding learning objectives, chapter summaries, and cross-references to related chapters.","why":"Brings the rest of the book to the same structural depth as the chapters expanded earlier the same day.","chapters":[["Biological Data Infrastructure","/foundations/data-infrastructure.html"],["Antibody and Biologic Design","/molecular/antibodies.html"],["Nucleic Acid and Genome Models","/molecular/genome-models.html"],["Cell Painting and Image-Based Phenotyping","/cells/cell-painting.html"],["Perturbation Prediction and Virtual Cells","/cells/virtual-cells.html"],["Target Identification and Prioritization","/therapeutics/targets.html"],["Robotic Lab Automation and Cloud Labs","/automation/cloud-labs.html"],["Benchmarks for Bio AI","/governance/benchmarks.html"]],"sources":[{"label":"Cimini et al., 2023","url":"https://doi.org/10.1038/s41596-023-00840-9","type":"paper"},{"label":"Carpenter et al., 2006","url":"https://doi.org/10.1186/gb-2006-7-10-r100","type":"paper"}]},{"category":"maintenance","title":"Behind-the-scenes maintenance","change":"Behind-the-scenes maintenance; no change to chapter content.","why":"","chapters":[],"sources":[]}]},
{"date":"2026-05-24","headline":"Six new chapters, six-part restructuring, and evidence anchors added across the handbook","entries":[{"category":"revision","title":"Front matter: orientation, scope, and citation guidance","change":"The Executive Summary, Preface, How to Read, Scope and Limitations, and How to Cite pages set out role-specific reading paths for seven reader types, explicit in-scope and out-of-scope sections, and citation formats in Suggested, APA, AMA, Vancouver, Chicago, and BibTeX styles.","why":"Gives readers clearer orientation on what the handbook covers, who it serves, how to read it by role, and how to cite it.","chapters":[["Executive Summary","/executive-summary.html"],["Preface","/preface.html"],["How to Read This Handbook","/how-to-read.html"],["Scope and Limitations","/scope-limitations.html"],["How to Cite This Handbook","/how-to-cite.html"]],"sources":[]},{"category":"content","title":"Glossary, Model Index, Case Studies, and References","change":"The Glossary, Model and Dataset Index, Case Studies, Chapter Summaries, and References pages carry term definitions, a canonical table of over 50 models and datasets, twelve applied case studies, and a summary of every chapter, with citation links given as DOI addresses.","why":"Gives readers verified, topic-organized reference tools for terminology, models, evidence-based cases, and citations across the handbook.","chapters":[["References","/references.html"],["Glossary","/appendices/glossary.html"],["Model and Dataset Index","/appendices/model-dataset-index.html"],["Case Studies","/appendices/case-studies.html"],["Quick Reference: All Chapter Summaries","/appendices/tldr-compilation.html"]],"sources":[]},{"category":"milestone","title":"Book restructured from eight parts to six","change":"The handbook's navigation and part structure are consolidated from eight nav items and eight book parts to seven nav items and six parts (for example, Therapeutic Discovery and Translation, Research Systems and Automation, Practice, Governance, and Futures), with part names updated across the Executive Summary, Foundations, How to Read, and Chapter Summaries pages.","why":"Reduces single-chapter top-level menu items and gives the book a clearer six-part structure.","chapters":[["Quick Reference: All Chapter Summaries","/appendices/tldr-compilation.html"],["Executive Summary","/executive-summary.html"],["AI for the Life Sciences","/foundations/ai-for-life-sciences.html"],["How to Read This Handbook","/how-to-read.html"],["Welcome","/"]],"sources":[]},{"category":"evidence","title":"Evidence anchors and primary citations added across most chapters","change":"Nearly every chapter gains new evidence-anchor tables or inline citations linking existing claims to primary sources, standards, and datasets, such as the NIST AI Risk Management Framework, model and dataset documentation standards, and benchmark and developability studies, adding 125 sources newly cited in the handbook.","why":"Ties existing claims throughout the handbook to citable primary sources, standards, and benchmark studies.","chapters":[["Evaluation Principles for Life Sciences AI","/foundations/evaluation-principles.html"],["Protein Design and Engineering","/molecular/protein-design.html"],["Single-Cell Foundation Models","/cells/single-cell-models.html"],["mRNA, RNA, and Vaccine Design","/therapeutics/vaccines.html"],["Synthetic Biology Design Tools","/automation/synbio-tools.html"],["Reproducibility and Open Science","/governance/reproducibility.html"],["Case Studies","/appendices/case-studies.html"],["References","/references.html"]],"sources":[{"label":"Leeman et al., 2024","url":"https://doi.org/10.1103/PRXEnergy.3.011002","type":"paper"},{"label":"FDA, Companion Diagnostics, 2026","url":"https://www.fda.gov/medical-devices/in-vitro-diagnostics/companion-diagnostics","type":"policy"},{"label":"NIH, Dual-Use Research of Concern Policy, 2024","url":"https://www.nih.gov/about-nih/who-we-are/nih-director/statements/statement-release-usg-policy-oversight-dual-use-research-concern-pathogens-enhanced-pandemic-potential","type":"policy"}]},{"category":"content","title":"Six new chapters added, spanning imaging, therapeutics, and biomanufacturing","change":"Adds six chapters: Knowledge Graphs and Literature AI, Histopathology AI, Microscopy and Cryo-EM AI, Drug Repurposing and Combination Therapy, Real-World Evidence and Biomarker AI, and Biomanufacturing, each with matching glossary terms, model index entries, and chapter summaries.","why":"Extends the handbook's coverage into the literature and knowledge-graph evidence layer, imaging modalities, drug repurposing, real-world data, and biomanufacturing.","chapters":[["Knowledge Graphs and Literature AI","/foundations/knowledge-and-literature.html"],["Histopathology AI","/cells/histopathology.html"],["Microscopy and Cryo-EM AI","/cells/microscopy-cryo-em.html"],["Drug Repurposing and Combination Therapy","/therapeutics/repurposing-combinations.html"],["Real-World Evidence and Biomarker AI","/therapeutics/real-world-evidence.html"],["AI for Biomanufacturing","/automation/biomanufacturing.html"]],"sources":[{"label":"Lee et al., 2019","url":"https://doi.org/10.1093/bioinformatics/btz682","type":"paper"},{"label":"Himmelstein et al., 2017","url":"https://doi.org/10.7554/elife.26726","type":"paper"},{"label":"Stringer et al., 2020","url":"https://doi.org/10.1038/s41592-020-01018-x","type":"paper"},{"label":"Subramanian et al., 2017","url":"https://doi.org/10.1016/j.cell.2017.10.049","type":"paper"},{"label":"FDA, Real-World Evidence Program, 2026","url":"https://www.fda.gov/science-research/science-and-research-special-topics/real-world-evidence","type":"policy"},{"label":"FDA, PAT Framework, 2004","url":"https://www.fda.gov/regulatory-information/search-fda-guidance-documents/pat-framework-innovative-pharmaceutical-development-manufacturing-and-quality-assurance","type":"policy"}]},{"category":"revision","title":"Welcome page and subtitle refined","change":"The Welcome page adopts the subtitle 'From Foundation Model to Functional Biology,' restores a disclaimer describing what the handbook does and does not cover, adds a line noting continuous updates as papers and models change, and simplifies the homepage's visual presentation of links to companion handbooks. The Preface removes a section on how the handbook was built.","why":"Sharpens the handbook's public framing and subtitle after the day's structural and content changes.","chapters":[["Welcome","/"],["Preface","/preface.html"],["Scope and Limitations","/scope-limitations.html"],["How to Cite This Handbook","/how-to-cite.html"]],"sources":[]}]},
{"date":"2026-05-25","headline":"Handbook restructured into six parts with nine new chapters and expanded evidence","entries":[{"category":"milestone","title":"Handbook restructured into six parts with nine new chapters","change":"Reorganizes the handbook into six parts and adds nine chapters: Systems Biology and Multiscale Modeling, Aging and Longevity Biology AI, Plant, Crop, and Agricultural AI, Environmental and Ecological AI, a new neuroscience chapter, Virtual Organisms and Digital Biology, Cell and Gene Therapy AI, Chemical Biology and Target Engagement, and Diagnostics and Biomarker Translation.","why":"Extends the handbook's scope from molecular and cellular AI into organismal, environmental, and translational domains that previously had no dedicated coverage.","chapters":[["Systems Biology and Multiscale Modeling","/cells/systems-biology.html"],["Aging and Longevity Biology AI","/organisms/aging-longevity.html"],["Plant, Crop, and Agricultural AI","/organisms/agriculture-ai.html"],["Environmental and Ecological AI","/organisms/environmental-ecology-ai.html"],["Neuroscience AI and Brain Foundation Models","/organisms/neuroscience-ai.html"],["Virtual Organisms and Digital Biology","/organisms/virtual-organisms.html"],["Cell and Gene Therapy AI","/therapeutics/cell-gene-therapy.html"],["Chemical Biology and Target Engagement","/therapeutics/chemical-biology.html"]],"sources":[{"label":"Badia-i-Mompel et al., 2023","url":"https://doi.org/10.1038/s41576-023-00618-5","type":"paper"},{"label":"Karr et al., 2012","url":"https://doi.org/10.1016/j.cell.2012.05.044","type":"paper"},{"label":"Wang et al., 2025","url":"https://doi.org/10.1038/s41586-025-08829-y","type":"paper"}]},{"category":"evidence","title":"Organismal and therapeutic methods, from genomic selection to BEELINE","change":"Eight chapters on agriculture, environmental ecology, systems biology, aging, neuroscience, virtual organisms, cell and gene therapy, and chemical biology cite primary studies including genomic selection (Meuwissen et al., 2001), plant breeding prediction (Crossa et al., 2017), SCENIC (Aibar et al., 2017), and the BEELINE benchmark (Pratapa et al., 2020).","why":"Lets readers trace AI methods in agriculture, ecology, systems biology, aging, neuroscience, and therapeutics to the primary studies behind them, from genomic selection to the BEELINE benchmark.","chapters":[["Plant, Crop, and Agricultural AI","/organisms/agriculture-ai.html"],["Environmental and Ecological AI","/organisms/environmental-ecology-ai.html"],["Systems Biology and Multiscale Modeling","/cells/systems-biology.html"],["Aging and Longevity Biology AI","/organisms/aging-longevity.html"],["Neuroscience AI and Brain Foundation Models","/organisms/neuroscience-ai.html"],["Virtual Organisms and Digital Biology","/organisms/virtual-organisms.html"],["Cell and Gene Therapy AI","/therapeutics/cell-gene-therapy.html"],["Chemical Biology and Target Engagement","/therapeutics/chemical-biology.html"]],"sources":[{"label":"Meuwissen et al., 2001","url":"https://doi.org/10.1093/genetics/157.4.1819","type":"paper"},{"label":"Crossa et al., 2017","url":"https://doi.org/10.1016/j.tplants.2017.08.011","type":"paper"},{"label":"Araus and Cairns, 2014","url":"https://doi.org/10.1016/j.tplants.2013.09.008","type":"paper"},{"label":"Aibar et al., 2017","url":"https://doi.org/10.1038/nmeth.4463","type":"paper"},{"label":"Pratapa et al., 2020","url":"https://doi.org/10.1038/s41592-019-0690-6","type":"paper"},{"label":"Kamimoto et al., 2023","url":"https://doi.org/10.1038/s41586-022-05688-9","type":"paper"}]},{"category":"evidence","title":"Cell-imaging and protein-design tools added to the glossary","change":"Adds glossary entries for Cellpose3, a restoration-aware cell-segmentation release (Stringer et al., 2025); COMPSS, a computational enzyme-design workflow (Johnson et al., 2024); and EVOLVEpro, a few-shot active-learning method for protein evolution (Jiang et al., 2025).","why":"Gives readers primary sources for tools referenced across the cell-imaging and protein-design chapters this update expanded.","chapters":[["Glossary","/appendices/glossary.html"],["Model and Dataset Index","/appendices/model-dataset-index.html"],["Microscopy and Cryo-EM AI","/cells/microscopy-cryo-em.html"],["Protein Design and Engineering","/molecular/protein-design.html"]],"sources":[{"label":"Stringer et al., 2025","url":"https://doi.org/10.1038/s41592-025-02595-5","type":"paper"},{"label":"Johnson et al., 2024","url":"https://doi.org/10.1038/s41587-024-02214-2","type":"paper"},{"label":"Jiang et al., 2025","url":"https://doi.org/10.1126/science.adr6006","type":"paper"}]},{"category":"evidence","title":"Glossary adds counterfactual and trial-emulation terms","change":"Adds glossary definitions for counterfactual response, neural optimal transport, Perturb-CITE-seq, and target trial emulation, the concepts behind virtual-cell and perturbation-prediction claims.","why":"Gives readers primary citations for the causal-inference concepts used across the perturbation and virtual-cell chapters.","chapters":[["Glossary","/appendices/glossary.html"]],"sources":[{"label":"Bunne et al., 2023","url":"https://doi.org/10.1038/s41592-023-01969-x","type":"paper"},{"label":"Frangieh et al., 2021","url":"https://doi.org/10.1038/s41588-021-00779-1","type":"paper"},{"label":"Hernán et al., 2022","url":"https://doi.org/10.1001/jama.2022.21383","type":"paper"}]},{"category":"revision","title":"The Big Picture: chapter summaries that lead with the central claim","change":"In roughly 30 chapters across foundations, molecular design, cells, automation, governance, and therapeutics, the Chapter Summary (TL;DR) begins with 'The Big Picture', which orients the reader to the chapter's central claim before the full text.","why":"Gives readers a consistent, more scannable orientation to each chapter's central claim before the full text.","chapters":[["Agentic Science Workflows","/automation/agentic-workflows.html"],["Perturbation Prediction and Virtual Cells","/cells/virtual-cells.html"],["Foundation Models for Biology","/foundations/foundation-models.html"],["Benchmarks for Bio AI","/governance/benchmarks.html"],["Protein Design and Engineering","/molecular/protein-design.html"],["Target Identification and Prioritization","/therapeutics/targets.html"],["Real-World Evidence and Biomarker AI","/therapeutics/real-world-evidence.html"],["Translational Evidence and Failure Modes","/therapeutics/translation-failures.html"]],"sources":[]},{"category":"revision","title":"Chapters reframed around a five-question utility standard","change":"Replaces the Demonstrated, Theoretical, and Beyond Current Capabilities headings with five standard questions, including what evidence would change the claim and what a researcher should do with it, across roughly 30 chapters.","why":"Gives readers a consistent, more actionable framework for judging capability claims in every chapter.","chapters":[["AI for the Life Sciences","/foundations/ai-for-life-sciences.html"],["Protein Structure Prediction","/molecular/protein-structure.html"],["History of AI in the Life Sciences","/foundations/history.html"],["Benchmarks for Bio AI","/governance/benchmarks.html"],["Single-Cell Foundation Models","/cells/single-cell-models.html"],["Target Identification and Prioritization","/therapeutics/targets.html"],["Virtual Organisms and Digital Biology","/organisms/virtual-organisms.html"],["Self-Driving Laboratories","/automation/self-driving-labs.html"]],"sources":[]},{"category":"revision","title":"Welcome page framing on foundation models tightened","change":"The Welcome page frames the handbook's scope around foundation models that predict protein structures, represent cells, and generate molecules, and presents the handbook as a field guide to the experimental work that remains; How to Cite carries the same description.","why":"Sharpens how the handbook introduces its scope and positioning to new readers.","chapters":[["Welcome","/"],["How to Cite This Handbook","/how-to-cite.html"]],"sources":[]},{"category":"accuracy","title":"Source verification","change":"Citations, figures, and links checked against their primary sources.","why":"","chapters":[],"sources":[]}]},
{"date":"2026-05-26","headline":"NIST screening evidence added; chapter framework renamed and voice tightened","entries":[{"category":"revision","title":"Chapter framework renamed to Field Guide; summaries condensed","change":"Renames the per-chapter orientation layer from 'Field Reference' to 'Field Guide' with a 'What would make this more promising?' question, and condenses the Chapter Summary (TL;DR) sections across roughly 35 chapters, cutting several hundred words from many of them.","why":"Simplifies chapter navigation and shortens summaries that had grown long after the previous day's expansion.","chapters":[["Agentic Science Workflows","/automation/agentic-workflows.html"],["Single-Cell Foundation Models","/cells/single-cell-models.html"],["AI for the Life Sciences","/foundations/ai-for-life-sciences.html"],["History of AI in the Life Sciences","/foundations/history.html"],["Protein Design and Engineering","/molecular/protein-design.html"],["Target Identification and Prioritization","/therapeutics/targets.html"],["Virtual Organisms and Digital Biology","/organisms/virtual-organisms.html"],["Benchmarks for Bio AI","/governance/benchmarks.html"]],"sources":[]},{"category":"evidence","title":"NIST biosecurity screening evaluation and pathology model findings","change":"Adds a blinded NIST inter-tool analysis finding baseline nucleic acid synthesis screening performance above 95% (Ikonomova et al., 2025; Laird et al., 2025) to the Synthetic Biology Tools and Information Hazards chapters, and a pathology foundation model fine-tuned for EGFR mutation prediction (Campanella et al., 2025) to the Histopathology AI chapter.","why":"Gives readers primary evidence for the safety-screening and pathology-translation claims these chapters make.","chapters":[["Synthetic Biology Design Tools","/automation/synbio-tools.html"],["Information Hazards in Capability Research","/governance/information-hazards.html"],["Histopathology AI","/cells/histopathology.html"]],"sources":[{"label":"Laird et al., 2025","url":"https://doi.org/10.1177/15356760251401228","type":"paper"},{"label":"Ikonomova et al., 2025","url":"https://www.nist.gov/publications/experimental-evaluation-ai%2Ddriven-protein-design-risks-using-safe-biological-proxies","type":"policy"},{"label":"Campanella et al., 2025","url":"https://doi.org/10.1038/s41591-025-03780-x","type":"paper"}]},{"category":"content","title":"Glossary expanded with about 40 new terms","change":"Adds roughly 40 new glossary entries, including A-Lab, AAV, analytical validity, antisense oligonucleotide, and base editing, each defining the term and noting where it applies in the handbook.","why":"Gives readers a fuller reference for terminology used across the handbook's molecular, cell, and translational chapters.","chapters":[["Glossary","/appendices/glossary.html"]],"sources":[]},{"category":"revision","title":"Start Here and Companion Handbooks on the Welcome page","change":"Tightens wording on the Welcome page, renaming sections to 'Start Here' and 'Companion Handbooks,' and in the Executive Summary, Foundations, How to Read, Preface, and Scope and Limitations pages.","why":"Continues sharpening the handbook's framing and reading guidance after the previous day's restructuring.","chapters":[["Welcome","/"],["Executive Summary","/executive-summary.html"],["AI for the Life Sciences","/foundations/ai-for-life-sciences.html"],["How to Read This Handbook","/how-to-read.html"],["Preface","/preface.html"],["Scope and Limitations","/scope-limitations.html"]],"sources":[]},{"category":"revision","title":"Handbook voice tightened across roughly 50 chapters","change":"Trims repeated phrasing and shortens chapter introductions throughout the handbook, and simplifies a few headings, for example renaming 'AI-Native Drug Discovery Companies' to 'Drug Discovery Companies Using AI' and 'Frontier Directions' to 'Open Technical Directions.'","why":"Reduces repetition and promotional-sounding phrasing left over from the week's rapid restructuring.","chapters":[["AI for the Life Sciences","/foundations/ai-for-life-sciences.html"],["Protein Structure Prediction","/molecular/protein-structure.html"],["Small Molecule Generation and ADMET","/therapeutics/small-molecules.html"],["Single-Cell Foundation Models","/cells/single-cell-models.html"],["Agentic Science Workflows","/automation/agentic-workflows.html"],["Information Hazards in Capability Research","/governance/information-hazards.html"],["Target Identification and Prioritization","/therapeutics/targets.html"],["Neuroscience AI and Brain Foundation Models","/organisms/neuroscience-ai.html"]],"sources":[]}]},
{"date":"2026-05-27","headline":"Behind-the-scenes maintenance","entries":[{"category":"maintenance","title":"Behind-the-scenes maintenance","change":"Behind-the-scenes maintenance; no change to chapter content.","why":"","chapters":[],"sources":[]}]},
{"date":"2026-06-02","headline":"Dangerous capability evaluation guidance: frontier safety frameworks and WMDP added","entries":[{"category":"developments","title":"OpenAI, Anthropic, and Google DeepMind biological capability safety frameworks","change":"Foundation Models for Biology notes that OpenAI's Preparedness Framework, Anthropic's Responsible Scaling Policy (ASL-3), and Google DeepMind's Frontier Safety Framework tie biological and CBRN capability thresholds to deployment mitigations, access controls, and model-weight security, adding that a gated model is not necessarily scientifically weaker.","why":"Gives readers primary sources for how frontier developers connect biological capability evaluation to release decisions.","chapters":[["Foundation Models for Biology","/foundations/foundation-models.html"]],"sources":[{"label":"OpenAI, Preparedness Framework, 2025","url":"https://openai.com/index/updating-our-preparedness-framework/","type":"announcement"},{"label":"Anthropic, Responsible Scaling Policy v3.0, 2026","url":"https://anthropic.com/responsible-scaling-policy/rsp-v3-0","type":"announcement"},{"label":"Google DeepMind, Frontier Safety Framework, 2026","url":"https://storage.googleapis.com/deepmind-media/DeepMind.com/Blog/strengthening-our-frontier-safety-framework/frontier-safety-framework_3-1.pdf","type":"announcement"}]},{"category":"evidence","title":"Dangerous-capability evaluation: WMDP, LAB-Bench, and biological uplift studies","change":"Benchmarks adds a section on safety and dangerous-capability evaluation, covering the WMDP hazardous-knowledge proxy benchmark, the LAB-Bench evaluation, and uplift studies; it cites RAND's red-team study finding no statistically significant difference in biological attack plan viability with or without LLM assistance, and OpenAI's early-warning study on LLM-aided biological threat creation.","why":"Distinguishes dangerous-capability evaluation, whether a model lowers the barrier to biological misuse, from benchmarks that measure beneficial scientific capability.","chapters":[["Benchmarks for Bio AI","/governance/benchmarks.html"]],"sources":[{"label":"Li et al., 2024, preprint","url":"https://doi.org/10.48550/arXiv.2403.03218","type":"preprint"},{"label":"Laurent et al., 2024, preprint","url":"https://doi.org/10.48550/arXiv.2407.10362","type":"preprint"},{"label":"Mouton et al., RAND, 2024","url":"https://doi.org/10.7249/RRA2977-2","type":"report"},{"label":"OpenAI, 2024","url":"https://openai.com/index/building-an-early-warning-system-for-llm-aided-biological-threat-creation/","type":"announcement"}]},{"category":"evidence","title":"Google Co-Scientist and FutureHouse Robin added to agentic science evidence","change":"Agentic Workflows and Self-Driving Labs cite Google Co-Scientist for biomedical hypothesis generation and FutureHouse Robin for lab-in-the-loop candidate discovery in dry age-related macular degeneration alongside ChemCrow, Coscientist, Virtual Lab, and CellVoyager as demonstrated systems, and reference a Nature editorial cautioning that autonomous research still requires human authorization for consequential steps.","why":"Grounds claims about autonomous research systems in the published record rather than vendor description, and marks the line between demonstrated and ungoverned autonomy.","chapters":[["Agentic Science Workflows","/automation/agentic-workflows.html"],["Self-Driving Laboratories","/automation/self-driving-labs.html"]],"sources":[{"label":"Gottweis et al., 2026","url":"https://doi.org/10.1038/s41586-026-10644-y","type":"paper"},{"label":"Ghareeb et al., 2026","url":"https://doi.org/10.1038/s41586-026-10652-y","type":"paper"},{"label":"Nature, editorial on AI scientists, 2026","url":"https://doi.org/10.1038/d41586-026-01551-3","type":"paper"}]}]},
{"date":"2026-06-09","headline":"Behind-the-scenes maintenance","entries":[{"category":"maintenance","title":"Behind-the-scenes maintenance","change":"Behind-the-scenes maintenance; no change to chapter content.","why":"","chapters":[],"sources":[]}]},
{"date":"2026-06-23","headline":"FDA clinical-trials AI pilot RFI; ProJenX digital twin comparator in ALS trial","entries":[{"category":"policy","title":"FDA request for information on AI-enabled early-phase trials pilot","change":"Clinical Trial AI for Translational Research cites FDA's April 2026 Federal Register request for information on an AI-enabled early-phase clinical-trials pilot program, which explores whether AI can improve trial efficiency, safety monitoring, dose-selection, and early go/no-go decisions while aligning with the NIST AI Risk Management Framework.","why":"Gives readers the primary regulatory source for FDA's exploration of AI-assisted early-phase trial design.","chapters":[["Clinical Trial AI for Translational Research","/therapeutics/clinical-trials.html"]],"sources":[{"label":"FDA, 2026","url":"https://www.federalregister.gov/documents/2026/04/29/2026-08281/ai-enabled-optimization-of-early-phase-clinical-trials-pilot-program-request-for-information","type":"policy"}]},{"category":"developments","title":"ProJenX and Unlearn digital twin comparator for ALS trial PRO-101","change":"The chapter adds ProJenX's PRO-101 Phase 1 ALS trial (ClinicalTrials.gov NCT05279755) as a boundary case, noting ProJenX and Unlearn's announced ALS Digital Twin Generator would estimate ALSFRS-R, slow vital capacity, and plasma neurofilament light trajectories as a model-based comparator, not a substitute for randomised efficacy evidence.","why":"Illustrates that the credibility of an AI-generated external comparator rests on prespecified context of use, endpoints, and uncertainty analysis rather than the digital twin label.","chapters":[["Clinical Trial AI for Translational Research","/therapeutics/clinical-trials.html"]],"sources":[{"label":"ClinicalTrials.gov, NCT05279755","url":"https://clinicaltrials.gov/study/NCT05279755","type":"other"},{"label":"ProJenX and Unlearn, September 2024","url":"https://projenx.com/projenx-and-unlearn-announce-partnership-to-augment-als-clinical-trial-pro-101-with-digital-twin-model/","type":"announcement"}]}]},
{"date":"2026-06-29","headline":"Behind-the-scenes maintenance","entries":[{"category":"maintenance","title":"Behind-the-scenes maintenance","change":"Behind-the-scenes maintenance; no change to chapter content.","why":"","chapters":[],"sources":[]}]},
{"date":"2026-07-02","headline":"Anthropic launches Claude Science, a vendor-reported beta AI workbench","entries":[{"category":"developments","title":"Claude Science, Anthropic's beta AI workbench for scientific research","change":"Foundations, Workforce and Compute, and Toolkit add Anthropic's June 2026 beta launch of Claude Science, an AI workbench linking literature search, data analysis, code execution, remote compute, and scientific software; the chapters label the evidence vendor-reported and give practical criteria (data boundary, exportable code, rerunnability) for evaluating it.","why":"Anthropic's own product announcement is the only current source for Claude Science; the chapters treat industrial workbench adoption as a signal to watch, not evidence of scientific productivity.","chapters":[["AI for the Life Sciences","/foundations/ai-for-life-sciences.html"],["Workforce, Compute, and Institutional Readiness","/governance/workforce-compute.html"],["Toolkit for AI-Augmented Bio Research","/practical/toolkit.html"]],"sources":[{"label":"Anthropic, June 2026","url":"https://www.anthropic.com/news/claude-science-ai-workbench","type":"announcement"}]}]},
{"date":"2026-07-04","headline":"Behind-the-scenes maintenance","entries":[{"category":"maintenance","title":"Behind-the-scenes maintenance","change":"Behind-the-scenes maintenance; no change to chapter content.","why":"","chapters":[],"sources":[]}]},
{"date":"2026-07-07","headline":"Source verification","entries":[{"category":"accuracy","title":"Source verification","change":"Citations, figures, and links checked against their primary sources.","why":"","chapters":[],"sources":[]}]},
{"date":"2026-07-13","headline":"Behind-the-scenes maintenance","entries":[{"category":"maintenance","title":"Behind-the-scenes maintenance","change":"Behind-the-scenes maintenance; no change to chapter content.","why":"","chapters":[],"sources":[]}]},
{"date":"2026-07-16","headline":"Executive Order 14292 on DURC/PEPP oversight; rentosertib reaches Phase 2a","entries":[{"category":"policy","title":"Executive Order 14292 and NIH's rescinded DURC/PEPP notice","change":"Executive Order 14292 directs agencies to revise or replace the 2024 U.S. government dual-use research policy, and NIH has rescinded its implementation notice NOT-OD-25-061. Information Hazards advises following current institutional and funder guidance for DURC and PEPP oversight.","why":"Keeps the handbook's dual-use oversight coverage aligned with the current regulatory status rather than the superseded 2024 policy notice.","chapters":[["Information Hazards in Capability Research","/governance/information-hazards.html"],["Quick Reference: All Chapter Summaries","/appendices/tldr-compilation.html"]],"sources":[{"label":"White House, Executive Order 14292, 2025","url":"https://www.whitehouse.gov/presidential-actions/2025/05/improving-the-safety-and-security-of-biological-research/","type":"policy"},{"label":"NIH, 2025","url":"https://grants.nih.gov/grants/guide/notice-files/NOT-OD-25-061.html","type":"policy"}]},{"category":"revision","title":"AI-discovered drug pipeline currency: rentosertib Phase 2a, Recursion, Insitro","change":"Case Studies, Executive Summary, and Small Molecule Generation and ADMET update Insilico Medicine's ISM001-055 (rentosertib) with a peer-reviewed Nature Medicine report on its randomized Phase 2a trial, alongside the earlier Nature Biotechnology discovery paper, and note Recursion's and Insitro's pipeline candidates; no AI-discovered small molecule has yet reached FDA or EMA approval.","why":"Reflects the trial's progression from discovery and Phase I to a randomized Phase 2a readout, while keeping the caveat that this is not registration-grade evidence.","chapters":[["Case Studies","/appendices/case-studies.html"],["Executive Summary","/executive-summary.html"],["Small Molecule Generation and ADMET","/therapeutics/small-molecules.html"]],"sources":[{"label":"Recursion, 2026","url":"https://www.recursion.com/pipeline","type":"other"}]},{"category":"accuracy","title":"Source verification","change":"Citations, figures, and links checked against their primary sources.","why":"","chapters":[],"sources":[]},{"category":"revision","title":"Virtual Cell Challenge results: no consistent outperformance of naive baselines","change":"Perturbation Prediction and Virtual Cells, Emerging Frontiers, and Benchmarks update to reflect the Virtual Cell Challenge's public results: models showed improvements on some perturbation-prediction metrics but did not consistently outperform naive baselines across all metrics, reframing the challenge as a shared benchmark rather than a paper-specific split.","why":"Gives readers the challenge's actual outcome rather than the earlier framing of the challenge as upcoming.","chapters":[["Perturbation Prediction and Virtual Cells","/cells/virtual-cells.html"],["Emerging Frontiers in AI for the Life Sciences","/future/emerging.html"],["Benchmarks for Bio AI","/governance/benchmarks.html"]],"sources":[]}]},
{"date":"2026-07-17","headline":"AI-designed antibiotics validated in mouse infection models","entries":[{"category":"evidence","title":"Antimicrobial discovery with experimental validation: SyntheMol-RL, ApexGO, CAMPER","change":"Small Molecule Generation and ADMET adds a section on antimicrobial discovery with experimental validation: SyntheMol-RL generated MRSA-active compounds tested in a mouse wound-infection model, ApexGO generated peptides evaluated in two Acinetobacter baumannii mouse infection models, and CAMPER prioritized membrane-targeting peptides tested against MRSA persisters in vitro and in mice.","why":"Adds experimentally validated evidence that generative and optimization methods can produce candidate antibiotics active against drug-resistant bacteria, not just in-silico benchmark performance.","chapters":[["Small Molecule Generation and ADMET","/therapeutics/small-molecules.html"]],"sources":[{"label":"Swanson et al., 2026","url":"https://doi.org/10.1038/s44320-026-00206-9","type":"paper"},{"label":"Torres et al., 2026","url":"https://doi.org/10.1038/s42256-026-01237-5","type":"paper"},{"label":"Shehadeh et al., 2026","url":"https://doi.org/10.1038/s41467-026-70348-9","type":"paper"},{"label":"SyntheMol-RL","url":"https://github.com/swansonk14/SyntheMol","type":"dataset"},{"label":"ApexGO","url":"https://github.com/Yimeng-Zeng/ApexGO","type":"dataset"},{"label":"CAMPER","url":"https://doi.org/10.5281/zenodo.17781367","type":"dataset"}]}]},
{"date":"2026-07-23","headline":"Source verification","entries":[{"category":"accuracy","title":"Source verification","change":"Citations, figures, and links checked against their primary sources.","why":"","chapters":[],"sources":[]}]},
{"date":"2026-07-26","headline":"Behind-the-scenes maintenance","entries":[{"category":"maintenance","title":"Behind-the-scenes maintenance","change":"Behind-the-scenes maintenance; no change to chapter content.","why":"","chapters":[],"sources":[]}]},
{"date":"2026-07-28","headline":"PBPK and QSP added as mature regulatory digital twins; benchmark uncertainty guidance","entries":[{"category":"evidence","title":"Statistical uncertainty in benchmark scores added to Evaluation Principles","change":"Evaluation Principles adds a section explaining that benchmark scores carry sampling uncertainty like any other estimate, recommending standard errors clustered by correlated test-case groups, paired comparison when models share a benchmark, and a power-analysis formula for the number of test cases needed to detect a claimed improvement.","why":"Completes the chapter's discussion of statistical significance testing for evaluation scores with a concrete method.","chapters":[["Evaluation Principles for Life Sciences AI","/foundations/evaluation-principles.html"]],"sources":[{"label":"Miller, 2024, preprint","url":"https://arxiv.org/abs/2411.00640","type":"preprint"}]},{"category":"content","title":"PBPK and QSP models: the field's most mature regulator-recognized digital twin","change":"Virtual Organisms and Digital Biology adds a section on PBPK and QSP models, reporting that 65 of 245 FDA new drug and biologics applications from 2020 to 2024 (26.5%) submitted a PBPK model as pivotal evidence, and covering the ASME V&V 40 credibility framework, FDA's 2023 guidance, MIDD, and the finalized ICH M15 guideline.","why":"Closes a coverage gap: PBPK, QSP, MIDD, and ASME V&V 40 had no prior mention despite PBPK's routine use as pivotal regulatory evidence.","chapters":[["Virtual Organisms and Digital Biology","/organisms/virtual-organisms.html"]],"sources":[{"label":"Sager et al., 2015","url":"https://doi.org/10.1124/dmd.115.065920","type":"paper"},{"label":"Li et al., 2025","url":"https://doi.org/10.3390/pharmaceutics17111413","type":"paper"},{"label":"Tegenge et al., 2025","url":"https://doi.org/10.1208/s12248-025-01155-1","type":"paper"},{"label":"FDA, considerations for AI in drug and biological product regulatory decisions, 2025","url":"https://www.fda.gov/regulatory-information/search-fda-guidance-documents/considerations-use-artificial-intelligence-support-regulatory-decision-making-drug-and-biological","type":"policy"},{"label":"FDA, Model-Informed Drug Development Paired Meeting Program","url":"https://www.fda.gov/drugs/development-resources/model-informed-drug-development-paired-meeting-program","type":"policy"},{"label":"FDA, M15 General Principles for Model-Informed Drug Development","url":"https://www.fda.gov/regulatory-information/search-fda-guidance-documents/m15-general-principles-model-informed-drug-development","type":"policy"}]},{"category":"content","title":"In silico clinical trial credibility: a test for AI model-matches-trial claims","change":"Virtual Organisms and Digital Biology adds an FAQ subsection on evaluating claims that a computational model matches clinical trial results, describing a hierarchical credibility-assessment workflow for in silico trials and citing its use in device submissions, pediatric trials, and randomized-trial augmentation with digital twins.","why":"Gives readers a generalizable test for vendor claims that a simulation reproduces trial outcomes, rather than a single accuracy figure.","chapters":[["Virtual Organisms and Digital Biology","/organisms/virtual-organisms.html"]],"sources":[{"label":"Pathmanathan et al., 2024","url":"https://doi.org/10.1371/journal.pcbi.1012289","type":"paper"},{"label":"Pammi et al., 2025","url":"https://doi.org/10.1016/j.landig.2025.01.007","type":"paper"},{"label":"Akbarialiabad et al., 2025","url":"https://doi.org/10.1038/s41540-025-00592-0","type":"paper"}]}]},
{"date":"2026-07-30","headline":"GPT-5.6 and Claude Opus 5 biosecurity classifications; five-layer evaluation framework","entries":[{"category":"developments","title":"Five-layer biosecurity evaluation framework; GPT-5.6 and Claude Opus 5 classifications","change":"Benchmarks, Executive Summary, and Chapter Summaries organize biosecurity evaluation into five layers, capability, uplift, safeguards, operational consequence, and lifecycle governance, and note that OpenAI treats all three GPT-5.6 models as High biological and chemical capability while Anthropic treats Claude Opus 5 as CB-1 but not CB-2 under ASL-3 protections.","why":"Distinguishes what a public benchmark can and cannot establish, and keeps developer capability classifications separate from independent uplift claims.","chapters":[["Benchmarks for Bio AI","/governance/benchmarks.html"],["Executive Summary","/executive-summary.html"],["Quick Reference: All Chapter Summaries","/appendices/tldr-compilation.html"]],"sources":[{"label":"OpenAI, GPT-5.6 system card, 2026","url":"https://deploymentsafety.openai.com/gpt-5-6","type":"announcement"},{"label":"Anthropic, Claude Opus 5 system card, 2026","url":"https://www.anthropic.com/claude-opus-5-system-card","type":"announcement"}]},{"category":"policy","title":"NIST AI 800-2 and 800-3 on benchmark evaluation validity","change":"Benchmarks cites NIST AI 800-2, an initial public draft on the validity, transparency, and reproducibility of automated benchmark evaluations, and NIST AI 800-3, which provides statistical methods for distinguishing fixed-benchmark performance from generalized task performance.","why":"Gives readers government standards for the statistical rigor benchmark claims should meet, separate from biosecurity-specific evaluation.","chapters":[["Benchmarks for Bio AI","/governance/benchmarks.html"]],"sources":[{"label":"NIST AI 800-2, initial public draft, 2026","url":"https://doi.org/10.6028/NIST.AI.800-2.ipd","type":"report"},{"label":"NIST AI 800-3, 2026","url":"https://doi.org/10.6028/NIST.AI.800-3","type":"report"}]},{"category":"evidence","title":"Uplift study evidence: substantial novice gains on bounded tasks, no full-workflow effect","change":"Benchmarks adds a 2026 preprint reporting substantial novice uplift on bounded digital biology tasks, and a separate preregistered randomized trial using mid-2025 models that found no significant difference in full-workflow completion for novices, though secondary analyses remained compatible with a possible modest benefit.","why":"Shows that uplift findings vary by task scope, bounded tasks versus full research workflows, and should not be read as a single capability ceiling.","chapters":[["Benchmarks for Bio AI","/governance/benchmarks.html"]],"sources":[{"label":"Hong et al., 2026, preprint","url":"https://arxiv.org/abs/2602.16703","type":"preprint"},{"label":"Zhang et al., 2026, preprint","url":"https://arxiv.org/abs/2602.23329","type":"preprint"}]},{"category":"evidence","title":"RFdiffusion3: all-atom protein design model, developer preprint","change":"Protein Design and Engineering and the Model and Dataset Index add RFdiffusion3, an all-atom diffusion model generating protein structures with ligands and nucleic acids; the developer preprint reports about tenfold faster generation than RFdiffusion2 with experimental validation limited to two selected design tasks, and releases inference code, training code, and model weights.","why":"Flags a new protein-design model as reported, developer-preprint evidence, since general performance across target classes is not yet established.","chapters":[["Protein Design and Engineering","/molecular/protein-design.html"],["Model and Dataset Index","/appendices/model-dataset-index.html"]],"sources":[{"label":"Butcher et al., 2025, preprint","url":"https://doi.org/10.1101/2025.09.18.676967","type":"preprint"},{"label":"RFdiffusion3, inference code, training code, and model weights","url":"https://github.com/RosettaCommons/foundry/blob/production/models/rfd3/README.md","type":"dataset"}]},{"category":"evidence","title":"Mutation-split choice inflates protein-language-model generalization claims","change":"Evaluation Principles adds a 2026 preprint analysis of 41 viral and 33 cellular deep-mutational-scanning datasets finding that pooled mutation splits let supervised protein-language-model predictors exploit site effects, inflating apparent generalization, while site-aware splits reduced performance and a site-mean baseline matched or outperformed supervised models on many datasets.","why":"Warns readers that mutational-effect benchmark results depend heavily on split design, though the finding is specific to single-substitution datasets.","chapters":[["Evaluation Principles for Life Sciences AI","/foundations/evaluation-principles.html"],["Foundation Models for Biology","/foundations/foundation-models.html"]],"sources":[{"label":"Vieira et al., 2026, preprint","url":"https://doi.org/10.64898/2026.03.08.710389","type":"preprint"}]},{"category":"evidence","title":"Retrosynthesis benchmark tools: Syntheseus, ChemCensor, and URSA-expert-2026","change":"Small Molecule Generation and ADMET adds Syntheseus, a controlled reevaluation that changed retrosynthesis model rankings under consistent settings, and a developer preprint proposing ChemCensor, a precedent-based plausibility score, and URSA-expert-2026, an expert-annotated out-of-distribution set of 100 novel targets; the chapter notes these test model comparison and plausibility, not route execution or experimental feasibility.","why":"Gives readers tools for retrosynthesis benchmark validity beyond exact-match accuracy, while flagging that neither scoring nor theoretical routes establish real synthesis feasibility.","chapters":[["Small Molecule Generation and ADMET","/therapeutics/small-molecules.html"]],"sources":[{"label":"Maziarz et al., 2025","url":"https://doi.org/10.1039/D4FD00093E","type":"paper"},{"label":"Zagribelnyy et al., 2026, preprint","url":"https://arxiv.org/abs/2602.03554","type":"preprint"}]},{"category":"evidence","title":"TargetBench 1.0: disease-specific evaluation of ranked drug targets","change":"Therapeutic Discovery and Translation adds TargetBench 1.0, an Insilico Medicine-led benchmark evaluating ranked target lists against historical clinical-stage targets and development-related evidence, noting that retrospective labels and text features can encode historical bias or temporal leakage, and that its code is proprietary.","why":"Gives readers a target-ranking benchmark while flagging risks specific to retrospective, text-based evaluation designs.","chapters":[["Target Identification and Prioritization","/therapeutics/targets.html"]],"sources":[{"label":"Leung et al., 2026","url":"https://doi.org/10.1038/s41598-026-47765-3","type":"paper"}]}]},
{"date":"2026-08-04","headline":"NIH DURC and PEPP notices aligned; agent security and evaluation-gap evidence added","entries":[{"category":"accuracy","title":"Source verification","change":"Citations, figures, and links checked against their primary sources.","why":"","chapters":[],"sources":[]},{"category":"policy","title":"NIST CAISI agent red-teaming results","change":"Information Hazards adds a summary noting that NIST's CAISI large-scale agent red-teaming competition found tool-use and agent-hijacking risk under adversarial testing, and states the results concern agent security rather than biological misuse capability.","why":"Distinguishes agent-hijacking findings from biological misuse capability so readers do not conflate the two kinds of risk.","chapters":[["Information Hazards in Capability Research","/governance/information-hazards.html"]],"sources":[{"label":"NIST CAISI, agent red-teaming competition insights, 2026","url":"https://www.nist.gov/blogs/caisi-research-blog/insights-ai-agent-security-large-scale-red-teaming-competition","type":"policy"}]},{"category":"policy","title":"International AI Safety Report on benchmark and real-world evidence","change":"Information Hazards now cites the 2026 International AI Safety Report to support its point that evaluation practice separates benchmark evidence from real-world biological outcomes.","why":"Adds an intergovernmental reference for the claim that benchmark performance does not by itself demonstrate real-world biological risk.","chapters":[["Information Hazards in Capability Research","/governance/information-hazards.html"]],"sources":[{"label":"International AI Safety Report, 2026","url":"https://internationalaisafetyreport.org/publication/international-ai-safety-report-2026","type":"policy"}]}]},
{"date":"2026-08-06","headline":"Source verification","entries":[{"category":"accuracy","title":"Source verification","change":"Citations, figures, and links checked against their primary sources.","why":"","chapters":[],"sources":[]}]},
{"date":"2026-08-07","headline":"Behind-the-scenes maintenance","entries":[{"category":"maintenance","title":"Behind-the-scenes maintenance","change":"Behind-the-scenes maintenance; no change to chapter content.","why":"","chapters":[],"sources":[]}]},
{"date":"2026-08-09","headline":"New genomic, single-cell, and protein design evidence across seven chapters","entries":[{"category":"evidence","title":"DeepDETAILS: cell-type signal reconstruction from bulk assays","change":"Microbiome and Multi-Omics AI and Variant Effect Prediction add DeepDETAILS, which reconstructs cell-type-specific regulatory signals from bulk assays such as PRO-cap, PRO-seq, and ChIP-seq to prioritize regulatory hypotheses in primary sclerosing cholangitis; the chapters note reconstructed signals are model-derived estimates, not direct single-cell measurements.","why":"Gives readers a new route to regulatory-variant hypotheses that still needs external functional validation.","chapters":[["Microbiome and Multi-Omics AI","/cells/multiomics.html"],["Variant Effect Prediction","/molecular/variant-effects.html"]],"sources":[{"label":"Yao et al., 2026","url":"https://doi.org/10.1038/s41587-026-03218-w","type":"paper"}]},{"category":"evidence","title":"HiCFoundation: pretrained 3D genome contact map representations","change":"Nucleic Acid and Genome Models adds HiCFoundation, evaluated across reproducibility assessment, resolution enhancement, loop detection, cross-species transfer, single-cell Hi-C analysis, and prediction of epigenomic activity from chromatin structure; Microbiome and Multi-Omics AI cross-references it for multi-omics use with single-cell and epigenomic endpoints.","why":"Adds a pretrained representation for 3D genome structure whose regulatory predictions remain hypotheses pending Hi-C-context validation.","chapters":[["Nucleic Acid and Genome Models","/molecular/genome-models.html"],["Microbiome and Multi-Omics AI","/cells/multiomics.html"]],"sources":[{"label":"Wang et al., 2026","url":"https://doi.org/10.1038/s41592-026-03097-8","type":"paper"}]},{"category":"evidence","title":"Measured pretraining scaling in single-cell foundation models","change":"Cells, Tissues, and Systems adds a section reporting that pretraining 400 models on up to 22.2 million cells across 6,400 model-task combinations found no clear data scaling law, and describes SIGnature, which linked the MS1 inflammatory program to Kawasaki disease and was followed by an in vitro test showing patient serum induced the phenotype.","why":"Tests whether scale alone improves single-cell models and shows one attribution-based hypothesis confirmed experimentally.","chapters":[["Single-Cell Foundation Models","/cells/single-cell-models.html"]],"sources":[{"label":"DenAdel et al., 2026","url":"https://doi.org/10.1038/s41592-026-03120-y","type":"paper"},{"label":"Gold et al., 2026","url":"https://doi.org/10.1038/s41587-026-03112-5","type":"paper"}]},{"category":"evidence","title":"Experimentally tested antibody and small-molecule binder design","change":"Protein Design and Engineering adds Germinal, which tested 43 to 101 antibody designs per antigen across four targets and reported functional binders including nanomolar-affinity designs, and Neural iterative selection-expansion, which reported binding for all four tested exatecan designs and five of six tested apixaban designs.","why":"Reports binder design results with tested-design denominators, though neither establishes therapeutic efficacy or manufacturability.","chapters":[["Protein Design and Engineering","/molecular/protein-design.html"]],"sources":[{"label":"Mille-Fragoso et al., 2026","url":"https://doi.org/10.1038/s41587-026-03187-0","type":"paper"},{"label":"Fry et al., 2026","url":"https://doi.org/10.1038/s41586-026-10670-w","type":"paper"}]},{"category":"evidence","title":"ContactSeek: structure-guided base editor specificity","change":"Cell and Gene Therapy AI adds ContactSeek, which combined AlphaFold 3 contact predictions with measured off-target data to prioritize residues for base-editor specificity, then tested selected changes in Cas9-TadA adenine base editors and a Cas12a-based cytosine editor.","why":"Adds a structure-model-guided workflow that still needs empirical off-target and genome-integrity testing before use.","chapters":[["Cell and Gene Therapy AI","/therapeutics/cell-gene-therapy.html"]],"sources":[{"label":"Meng et al., 2026","url":"https://doi.org/10.1038/s41586-026-10794-z","type":"paper"}]},{"category":"evidence","title":"AI-nominated Parkinson disease targets tested in mouse models","change":"Therapeutic Discovery and Translation adds a preclinical study in which XunZi nominated CHK2 and IRAK4 as candidate targets in Parkinson disease models, and pharmacological or genetic Chk2 inhibition reduced dopaminergic neuron loss and motor deficits in mice.","why":"Shows AI-assisted target hypothesis generation followed by experimental validation, short of human disease modification evidence.","chapters":[["Target Identification and Prioritization","/therapeutics/targets.html"]],"sources":[{"label":"Huang et al., 2026","url":"https://doi.org/10.1038/s41551-026-01769-6","type":"paper"}]},{"category":"revision","title":"Protein structure prediction entry reorganized","change":"Molecular Discovery and Design replaces its Field Guide heading with a direct explanation of protein structure prediction, restating that AlphaFold 2 and 3 and open systems such as Boltz and Chai produce structural hypotheses whose confidence metrics do not establish function, mechanism, or binding affinity.","why":"Clarifies what structure prediction shows and does not show, right where readers look for a definition.","chapters":[["Protein Structure Prediction","/molecular/protein-structure.html"]],"sources":[]}]},
{"date":"2026-08-11","headline":"New medical-countermeasure readiness section in vaccine design chapter","entries":[{"category":"content","title":"New section: vaccine design to medical-countermeasure readiness","change":"mRNA, RNA, and Vaccine Design adds a section arguing that faster antigen or sequence design alone does not establish end-to-end medical-countermeasure readiness, citing CEPI's 100 Days Mission, a BARDA-authored review of COVID-19 countermeasure development, a 2026 commentary mapping AI chokepoints across discovery, manufacturing, trials, and supply, CEPI's planned Pandemic Preparedness Engine, and joint FDA-EMA good AI practice principles.","why":"Separates faster computational design from the regulatory, manufacturing, and supply readiness a real pandemic response requires.","chapters":[["mRNA, RNA, and Vaccine Design","/therapeutics/vaccines.html"]],"sources":[{"label":"CEPI, 100 Days Mission","url":"https://cepi.net/100-days-mission","type":"announcement"},{"label":"Johnson et al., 2022","url":"https://doi.org/10.1080/21645515.2022.2129930","type":"paper"},{"label":"Adalja et al., 2026","url":"https://doi.org/10.1093/ofid/ofag108","type":"paper"},{"label":"CEPI, Pandemic Preparedness Engine","url":"https://cepi.net/artificial-intelligence/","type":"announcement"},{"label":"FDA and EMA, good AI practice principles","url":"https://www.fda.gov/about-fda/artificial-intelligence-drug-development/guiding-principles-good-ai-practice-drug-development","type":"policy"}]}]},
{"date":"2026-08-12","headline":"Handbook-wide heading restructuring; new high-risk life sciences research policy","entries":[{"category":"revision","title":"Capability-assessment headings restructured across the handbook","change":"Most chapters replace question-style headings (What is demonstrated?, What is theoretical?) with direct section names: Demonstrated capability, Theoretical capability, Beyond current capability, Evidence that would change the assessment, and Implications for research and program decisions, alongside updated evidence within each section.","why":"Gives each chapter's capability assessment a plain declarative structure instead of a question-and-answer format.","chapters":[["AI for the Life Sciences","/foundations/ai-for-life-sciences.html"],["Protein Structure Prediction","/molecular/protein-structure.html"],["Single-Cell Foundation Models","/cells/single-cell-models.html"],["Self-Driving Laboratories","/automation/self-driving-labs.html"],["Information Hazards in Capability Research","/governance/information-hazards.html"],["Small Molecule Generation and ADMET","/therapeutics/small-molecules.html"],["Virtual Organisms and Digital Biology","/organisms/virtual-organisms.html"],["Evaluation Principles for Life Sciences AI","/foundations/evaluation-principles.html"]],"sources":[]},{"category":"policy","title":"USG policy for stopping high-risk life sciences research","change":"Glossary now cites the July 2026 USG Policy for Stopping High-Risk Life Sciences Research and NIH notice NOT-OD-26-101 for the dual-use research of concern definition, replacing citations to the 2024 NIH statement, a 2025 White House action, and NIH notice NOT-OD-25-061.","why":"Reflects a new July 2026 government policy and NIH notice superseding the prior dual-use research oversight notices.","chapters":[["Glossary","/appendices/glossary.html"]],"sources":[{"label":"White House, USG Policy for Stopping High-Risk Life Sciences Research, July 2026","url":"https://www.whitehouse.gov/wp-content/uploads/2026/07/USG-Policy-for-Stopping-High-Risk-Life-Sciences-Research_July-2026.pdf","type":"policy"},{"label":"NIH, NOT-OD-26-101, 2026","url":"https://grants.nih.gov/grants/guide/notice-files/NOT-OD-26-101.html","type":"policy"}]},{"category":"accuracy","title":"Source verification","change":"Citations, figures, and links checked against their primary sources.","why":"","chapters":[],"sources":[]},{"category":"evidence","title":"AlphaFold 3 primary citation added to case studies","change":"Case Studies now cites Abramson et al., 2024 as the primary source for AlphaFold 3, alongside the restructured protein structure prediction case.","why":"Gives readers a direct citation for AlphaFold 3 rather than only a narrative description.","chapters":[["Case Studies","/appendices/case-studies.html"]],"sources":[{"label":"Abramson et al., 2024","url":"https://doi.org/10.1038/s41586-024-08416-7","type":"paper"}]}]},
{"date":"2026-08-15","headline":"Self-driving labs and protein structure prediction framing aligned","entries":[{"category":"revision","title":"Self-driving labs framing aligned with bounded evidence","change":"Self-Driving Labs restates that evidence spans closed-loop chemistry and materials systems while current biology examples remain lab-in-the-loop rather than autonomous robotic loops, citing Coscientist, Virtual Lab, Robin, and A-Lab alongside published critiques of A-Lab's novelty claims.","why":"Keeps the bounded-optimization versus general-autonomous-discovery distinction consistent with the cited evidence.","chapters":[["Self-Driving Laboratories","/automation/self-driving-labs.html"]],"sources":[]},{"category":"revision","title":"Protein structure prediction framing aligned across chapters","change":"Molecular Discovery and Design restates AlphaFold 2's routine role since CASP14, notes the AlphaFold Protein Structure Database now covers over 214 million predicted structures, and frames AlphaFold 3, Boltz-1, and Chai-1 alongside the questions a prediction still leaves for experiments to answer.","why":"Keeps the structure-prediction chapter consistent about what a prediction shows and what still needs experimental proof.","chapters":[["Protein Structure Prediction","/molecular/protein-structure.html"]],"sources":[]}]},
{"date":"2026-08-16","headline":"Evidence-to-claim framework and reading path added; Nature Methods on open AI","entries":[{"category":"content","title":"Evidence-to-claim matching table and decision-first reading path","change":"Evaluation Principles adds a table matching claim types, such as prediction versus experimental benefit, to the evidence design each requires; How to Read adds a decision-first reading path; the Preface states the book distinguishes predictions, benchmark results, validated findings, and decision changes.","why":"Gives readers explicit frameworks for judging what a result does and does not establish, and for navigating the handbook by decision rather than model name.","chapters":[["Evaluation Principles for Life Sciences AI","/foundations/evaluation-principles.html"],["How to Read This Handbook","/how-to-read.html"],["Preface","/preface.html"]],"sources":[]},{"category":"content","title":"Timeline of AI in biology added to History chapter","change":"Adds a milestone timeline from 1970 Needleman-Wunsch sequence alignment through BLAST, profile-HMM methods, DeepBind and DeepSEA, and DeepVariant, describing later systems as changing scale and reach while retaining earlier databases and evaluation practices.","why":"Shows readers that current systems built cumulatively on established methods rather than replacing the prior scientific record.","chapters":[["History of AI in the Life Sciences","/foundations/history.html"]],"sources":[]},{"category":"content","title":"Tool-adoption record and comparative trial guidance","change":"Adds a tool-adoption record template covering workload and owner, decision and endpoint, data boundaries, candidate tool, evidence basis, integration path, cost, and acceptance and exit tests, plus a guide for running a comparative tool trial.","why":"Gives research teams a structured, durable basis for evaluating and reviewing tools instead of relying on feature lists.","chapters":[["Toolkit for AI-Augmented Bio Research","/practical/toolkit.html"]],"sources":[]},{"category":"evidence","title":"Nature Methods Perspective on open and sustainable AI recommendations","change":"Reproducibility cites a Nature Methods Perspective mapping open and sustainable AI recommendations across the life-sciences ecosystem, covering data and model documentation, archival infrastructure, and environmental considerations, and notes it is an implementation resource, not proof that a checklist ensures reproducibility.","why":"Gives the chapter a primary source for openness and sustainability guidance spanning data, code, weights, and documentation.","chapters":[["Reproducibility and Open Science","/governance/reproducibility.html"],["References","/references.html"]],"sources":[{"label":"Farrell et al., 2026","url":"https://doi.org/10.1038/s41592-026-03037-6","type":"paper"}]}]},
{"date":"2026-08-18","headline":"NIST agent-evaluation draft; post-training data lineage and model-collapse evidence","entries":[{"category":"policy","title":"NIST draft standard treats agent settings as part of evaluation protocol","change":"Agentic Workflows adds a Configured-System Evaluation section describing the evaluated object as the full configured agent system, not the model alone, following a NIST initial public draft that calls for explicit objectives, reproducible runs, uncertainty analysis, and qualified reporting.","why":"Gives readers a draft government standard for evaluating agent systems rather than model names alone.","chapters":[["Agentic Science Workflows","/automation/agentic-workflows.html"]],"sources":[{"label":"NIST AI 800-2, initial public draft, 2026","url":"https://doi.org/10.6028/NIST.AI.800-2.ipd","type":"policy"}]},{"category":"evidence","title":"Agent evaluation benchmarks: repeated-trial and judge-bias findings","change":"The same section cites a study showing final-state and repeated-trial checks in customer-service benchmarks exposed failures that single conversational scores missed, and a study showing human and LLM judges are both susceptible to multiple forms of judgment bias, noting neither establishes biological performance.","why":"Supports the chapter's call for outcome, process, and safety checks beyond a single benchmark score, while flagging that these findings come from non-biological domains.","chapters":[["Agentic Science Workflows","/automation/agentic-workflows.html"]],"sources":[{"label":"Yao et al., 2025","url":"https://proceedings.iclr.cc/paper_files/paper/2025/hash/1b126cc38b8638e07bef37e7b2bb72bf-Abstract-Conference.html","type":"paper"},{"label":"Chen et al., 2024","url":"https://doi.org/10.18653/v1/2024.emnlp-main.474","type":"paper"}]},{"category":"evidence","title":"Post-training data lineage: Tulu 3, RLAIF, and model collapse evidence","change":"Biological Data Infrastructure adds a Post-Training Dataset Card and cites an open recipe exposing post-training data types and decontamination steps, a study showing AI-feedback reinforcement learning matched human-feedback results in summarization and dialogue, and a study showing indiscriminate recursive training on model-generated data can lose distribution tails.","why":"Grounds the chapter's guidance that post-training data types, generators, and evaluators must be tracked as separate provenance, not treated as interchangeable or as independent ground truth.","chapters":[["Biological Data Infrastructure","/foundations/data-infrastructure.html"]],"sources":[{"label":"Lambert et al., 2025","url":"https://openreview.net/forum?id=i1uGbfHHpH","type":"preprint"},{"label":"Lee et al., 2024","url":"https://proceedings.mlr.press/v235/lee24t.html","type":"paper"},{"label":"Shumailov et al., 2024","url":"https://doi.org/10.1038/s41586-024-07566-y","type":"paper"}]}]},
{"date":"2026-08-19","headline":"Field Guide sections added to History, Emerging Frontiers, and Protein Structure","entries":[{"category":"content","title":"Field Guide sections added to History, Emerging Frontiers, and Protein Structure","change":"History, Emerging Frontiers, and Molecular Discovery and Design each add a five-question Field Guide (what the field solves, the core idea, current state, what remains open, why it matters); the structure chapter now frames AlphaFold 2's single-chain accuracy against AlphaFold 3, Boltz, and Chai's extension to complexes.","why":"Applies one evidence-tiered framework across chapters so readers can gauge a claim's evidence level before acting on it.","chapters":[["History of AI in the Life Sciences","/foundations/history.html"],["Emerging Frontiers in AI for the Life Sciences","/future/emerging.html"],["Protein Structure Prediction","/molecular/protein-structure.html"]],"sources":[]}]},
{"date":"2026-08-20","headline":"New Start Here page adds reading paths; Welcome reframes translation as bottleneck","entries":[{"category":"content","title":"New Start Here page with reading paths","change":"Adds a new Start Here page offering three reading paths (new to life sciences AI, evaluating a model or vendor, exploring a research domain) and a browse list of the handbook's parts; the Welcome page introduces the handbook by framing translation, not the model, as the current bottleneck.","why":"Gives new readers a short, direct entry point into the handbook's reading paths and structure.","chapters":[["Start Here","/start.html"],["Welcome","/"]],"sources":[]}]},
{"date":"2026-08-21","headline":"Anthropic reports Claude-designed protein binders; clinical-trial AI evidence expanded","entries":[{"category":"developments","title":"Anthropic reports Claude-designed de novo protein binders","change":"Protein Design and Engineering adds Anthropic's reported binder campaign: Claude (Opus 4.8 and Mythos Preview), with a human-expert prompt and public design and folding models, produced de novo binders against 14 of 15 targets, independently produced and tested by Adaptyv Bio and Twist Bioscience, with reported hit rates of 22 to 35% against a field-typical 10 to 15%.","why":"Gives readers a primary company report of a large de novo binder campaign, while noting it is company-reported, not peer-reviewed, and the binders are not drugs.","chapters":[["Protein Design and Engineering","/molecular/protein-design.html"]],"sources":[{"label":"Anthropic, August 2026","url":"https://www.anthropic.com/research/Claude-accelerates-protein-design","type":"announcement"}]},{"category":"revision","title":"Clinical trial AI chapter reframed around context of use","change":"The chapter's opening now sets a context of use before selecting metrics, citing FDA's reported submission experience with AI components in drug development from 2016 to 2023, EMA's adopted reflection paper on AI across the medicinal product lifecycle, and a Nature Medicine review situating trial applications within the broader drug-development pipeline.","why":"Grounds the chapter's evidentiary standards in FDA and EMA frameworks and situates trial AI within the broader drug-development pipeline before it evaluates specific tools.","chapters":[["Clinical Trial AI for Translational Research","/therapeutics/clinical-trials.html"]],"sources":[]},{"category":"evidence","title":"Trial matching and enrollment evidence across five studies","change":"The new Matching evidence from retrieval to enrollment section adds five studies spanning candidate-trial matching, randomized prescreening and clinician-notification trials, and multisite eligibility classification, each showing evidence at one step, such as retrieval or screening, does not establish performance at another, such as enrollment.","why":"Gives the chapter primary evidence that trial-matching gains at one step, such as faster screening, do not automatically translate into gains at enrollment or outcomes.","chapters":[["Clinical Trial AI for Translational Research","/therapeutics/clinical-trials.html"]],"sources":[{"label":"Gueguen et al., 2025","url":"https://doi.org/10.1038/s41698-025-00806-y","type":"paper"},{"label":"Unlu et al., 2025","url":"https://doi.org/10.1001/jama.2024.28047","type":"paper"},{"label":"Mazor et al., 2025","url":"https://doi.org/10.1001/jamanetworkopen.2025.2013","type":"paper"},{"label":"Callies et al., 2025","url":"https://doi.org/10.1038/s43856-025-01256-0","type":"paper"},{"label":"Syed et al., 2026","url":"https://doi.org/10.1093/jamia/ocag006","type":"paper"}]},{"category":"evidence","title":"Trial Outcome Prediction and Clinical Trial Outcome benchmark datasets","change":"The new Trial outcome prediction and benchmark validity section adds the Trial Outcome Prediction benchmark (17,538 trials, retrospective phase-specific prediction) and the Clinical Trial Outcome dataset (about 125,000 drug and biologic trial records), noting the original benchmark authors flagged look-ahead bias as a validity threat.","why":"Gives readers the primary benchmark datasets behind trial-outcome-prediction claims and flags a known validity threat for future evaluations.","chapters":[["Clinical Trial AI for Translational Research","/therapeutics/clinical-trials.html"]],"sources":[{"label":"Fu et al., 2022","url":"https://doi.org/10.1016/j.patter.2022.100445","type":"paper"},{"label":"Gao et al., 2026","url":"https://doi.org/10.1038/s44360-026-00081-6","type":"paper"}]},{"category":"evidence","title":"Nature Medicine Perspective proposes AIDO multiscale foundation models","change":"Virtual Organisms and Digital Biology cites a Nature Medicine Perspective proposing AIDO, a modular system of multiscale foundation models spanning molecules to individuals, describing it as a research agenda in the same class as the virtual-cell roadmap, not evidence of an executable digital organism.","why":"Gives the chapter a primary source for the AIDO proposal while flagging the authors' financial interest in GenBio AI and that the article does not demonstrate a working digital organism.","chapters":[["Virtual Organisms and Digital Biology","/organisms/virtual-organisms.html"]],"sources":[{"label":"Song et al., 2026","url":"https://doi.org/10.1038/s41591-026-04595-0","type":"paper"}]}]},
{"date":"2026-08-23","headline":"New Evaluation Workflows chapter and worksheets; handbook DOI and archival PDF published","entries":[{"category":"content","title":"New Evaluation Workflows chapter with downloadable worksheets","change":"Adds a new Evaluation Workflows chapter with a universal decision record and five downloadable CSV worksheets covering configured-system evaluation, tool-versus-process comparison, single-cell model evaluation, self-driving-lab audits, and clinical-trial AI validation.","why":"Turns the handbook's evidence principles into repeatable, version-controllable worksheets teams can use without new software.","chapters":[["Practical Evaluation Workflows and Worksheets","/practical/evaluation-workflows.html"]],"sources":[]},{"category":"content","title":"Decision-record frameworks added to four evaluation chapters","change":"Self-Driving Labs, Cells/Tissues/Systems, Evaluation Principles, and Clinical Trial AI each add a decision-record subsection: classifying automation versus a closed autonomous loop, a model-adoption path for single-cell foundation models, a decision-ready evaluation protocol, and a decision chain mapping trial functions to immediate and downstream endpoints.","why":"Pairs each chapter with a structured record so a claim, such as autonomy or adoption, can be checked against explicit fields rather than a single score.","chapters":[["Self-Driving Laboratories","/automation/self-driving-labs.html"],["Single-Cell Foundation Models","/cells/single-cell-models.html"],["Evaluation Principles for Life Sciences AI","/foundations/evaluation-principles.html"],["Clinical Trial AI for Translational Research","/therapeutics/clinical-trials.html"]],"sources":[]},{"category":"content","title":"Clinical trial AI models added to Model and Dataset Index","change":"The Model and Dataset Index adds a Clinical Trial AI and Evidence Generation table listing TrialGPT, HINT, and the Clinical Trial Outcome dataset with their main use and evidence tier.","why":"Gives readers a quick-reference entry point to clinical-trial AI tools and datasets already covered in the chapter.","chapters":[["Model and Dataset Index","/appendices/model-dataset-index.html"]],"sources":[]},{"category":"milestone","title":"Handbook DOI and archival PDF edition published","change":"Publishes DOI 10.5281/zenodo.22073136 for the handbook, resolving to a newly prepared archival PDF edition; several chapters (Self-Driving Labs, Cells/Tissues/Systems, Evaluation Principles, Evaluation Workflows, Clinical Trial AI) replace decision tables with equivalent diagrams for that edition, and How to Cite adds citation formats using the DOI.","why":"Gives the handbook a permanent, citable identifier and a fixed archival PDF edition, distinct from the continuously updated website.","chapters":[["How to Cite This Handbook","/how-to-cite.html"],["Guidance for AI Agents","/for-ai-agents.html"],["Welcome","/"],["Self-Driving Laboratories","/automation/self-driving-labs.html"],["Single-Cell Foundation Models","/cells/single-cell-models.html"],["Evaluation Principles for Life Sciences AI","/foundations/evaluation-principles.html"],["Practical Evaluation Workflows and Worksheets","/practical/evaluation-workflows.html"],["Clinical Trial AI for Translational Research","/therapeutics/clinical-trials.html"]],"sources":[]}]},
{"date":"2026-08-25","headline":"Behind-the-scenes maintenance","entries":[{"category":"maintenance","title":"Behind-the-scenes maintenance","change":"Behind-the-scenes maintenance; no change to chapter content.","why":"","chapters":[],"sources":[]}]},
{"date":"2026-09-02","headline":"29 evidence bounds added across molecular, cellular, and therapeutic chapters","entries":[{"category":"evidence","title":"Protein design and engineering: four new evidence bounds","change":"Protein Design and Engineering adds four bounds: EvoMax for wet-lab nuclease engineering, ProteinDPO aligning ESM-IF1 inverse folding to experimental stability data, DeepSCan mapping sequence to cell-surface translocation for display design, and a mammalian-display screen of thousands of AI-designed minibinders against PD-L1, CD276, and VTCN1.","why":"Gives the chapter concrete wet-lab-validated bounds on what current design and scoring methods demonstrate, distinct from computational predictions alone.","chapters":[["Protein Design and Engineering","/molecular/protein-design.html"]],"sources":[{"label":"Wan et al., 2026","url":"https://doi.org/10.1038/s41587-026-03272-4","type":"paper"},{"label":"Widatalla et al., 2026","url":"https://doi.org/10.1038/s41592-026-03137-3","type":"paper"},{"label":"Fang et al., 2026","url":"https://doi.org/10.1038/s41587-026-03144-x","type":"paper"},{"label":"Broske et al., 2026","url":"https://doi.org/10.1038/s41467-026-76760-5","type":"paper"}]},{"category":"evidence","title":"ESM-3 dynamics readouts: Dyna-1 and ESMDynamic bounds","change":"Molecular Discovery and Design adds two ESM-3 dynamics readouts: Dyna-1, trained on missing NMR assignments to mark microsecond-to-millisecond motion, and ESMDynamic, a dynamic-contact-map bound, both framed as readouts on top of ESM-3 rather than new base models.","why":"Shows readers how new dynamics-prediction claims relate to the existing ESM-3 base model rather than replacing it.","chapters":[["Protein Structure Prediction","/molecular/protein-structure.html"]],"sources":[{"label":"Wayment-Steele et al., 2026","url":"https://doi.org/10.1038/s41586-026-10989-4","type":"paper"},{"label":"Kleiman et al., 2026","url":"https://doi.org/10.1038/s41467-026-76361-2","type":"paper"}]},{"category":"evidence","title":"RNA pseudoknot design and methylation-scoring bounds","change":"Nucleic Acid and Genome Models adds a structure-creation bound on specified RNA pseudoknots, distinguished from viral-function claims, and Melody, which scores locus-specific DNA methylation from 10-kb genomic sequence across 39 human tissues.","why":"Gives the chapter bounded examples of RNA structure design and methylation scoring rather than broader genome-regulatory claims.","chapters":[["Nucleic Acid and Genome Models","/molecular/genome-models.html"]],"sources":[{"label":"Townley et al., 2026","url":"https://doi.org/10.1126/science.aeg6829","type":"paper"},{"label":"Jin et al., 2026","url":"https://doi.org/10.1038/s41467-026-76744-5","type":"paper"}]},{"category":"evidence","title":"Antibody-discovery and rare-variant-effect benchmark bounds","change":"Antibody and Biologic Design adds AIntibody, a blinded prospective benchmark testing 511 designed or predicted antibodies from 29 organizations; Variant Effect Prediction adds RareEffect, an empirical-Bayes method estimating gene-level and variant-level rare-variant effect sizes.","why":"Gives both chapters blinded or statistically grounded benchmarks rather than self-reported design or scoring claims.","chapters":[["Antibody and Biologic Design","/molecular/antibodies.html"],["Variant Effect Prediction","/molecular/variant-effects.html"]],"sources":[{"label":"Erasmus et al., 2026","url":"https://doi.org/10.1038/s41587-026-03238-6","type":"paper"},{"label":"Nam et al., 2026","url":"https://doi.org/10.1038/s41588-026-02705-9","type":"paper"}]},{"category":"evidence","title":"Histopathology AI: four new benchmark and model bounds","change":"Histopathology AI adds nnMIL, a slide-level aggregator over existing pathology foundation-model embeddings; ALPaCA, a slide-level vision-language framework trained on 35,913 whole-slide images; TRICARE, which scores risk across 2D levels inside 3D pathology volumes; and ConceptCLIP, a concept-aligned vision-language model pretrained on 23 million image-text-concept triplets.","why":"Extends the chapter's evidence base across slide aggregation, visual question answering, 3D triage, and concept-aligned representation learning.","chapters":[["Histopathology AI","/cells/histopathology.html"]],"sources":[{"label":"Luo et al., 2026","url":"https://doi.org/10.1038/s41551-026-01767-8","type":"paper"},{"label":"Gao et al., 2026 (ALPaCA)","url":"https://doi.org/10.1038/s41467-026-76372-z","type":"paper"},{"label":"Gao et al., 2026 (TRICARE)","url":"https://doi.org/10.1038/s41551-026-01760-1","type":"paper"},{"label":"Nie et al., 2026","url":"https://doi.org/10.1038/s41551-026-01764-x","type":"paper"}]},{"category":"evidence","title":"Cell imaging and spatial-omics: six new evidence bounds","change":"Adds VirTues (spatial-proteomics cohort bounds), MOSS (an LLM-assisted toolkit existence proof), a cryo-ET phantom with ground-truth annotations for six molecular species, scE2TM (an interpretable topic-model embedding), Bonsai (tree-based visualization with improved nearest-neighbor recovery), and a Cell Painting assay pairing PMA activation with 8,387 compounds.","why":"Extends chapters covering spatial, imaging, and single-cell methods with named, source-grounded bounds on what each approach demonstrates.","chapters":[["Spatial Omics and Tissue Models","/cells/spatial-omics.html"],["Microscopy and Cryo-EM AI","/cells/microscopy-cryo-em.html"],["Single-Cell Foundation Models","/cells/single-cell-models.html"],["Cell Painting and Image-Based Phenotyping","/cells/cell-painting.html"],["Toolkit for AI-Augmented Bio Research","/practical/toolkit.html"]],"sources":[{"label":"Wenckstern et al., 2026","url":"https://doi.org/10.1038/s41586-026-10884-y","type":"paper"},{"label":"Medina and Kornfeld, 2026","url":"https://doi.org/10.1038/s41592-026-03210-x","type":"paper"},{"label":"Peck et al., 2025","url":"https://doi.org/10.1038/s41592-025-02800-5","type":"paper"},{"label":"Chen et al., 2026 (scE2TM)","url":"https://doi.org/10.1038/s41467-026-76825-5","type":"paper"},{"label":"de Groot et al., 2026","url":"https://doi.org/10.1038/s41587-026-03220-2","type":"paper"},{"label":"Zietek et al., 2026","url":"https://doi.org/10.1038/s41467-026-76252-6","type":"paper"}]},{"category":"evidence","title":"Therapeutics: four new evidence bounds across three chapters","change":"Chemical Biology and Target Engagement adds VITAL, which scores peptide-protein interactions and affinity from language-model embeddings, and a MaSIF-mimicry surface-query screen of CRBN binding; Cell and Gene Therapy AI adds OptiPrime for prime-editing efficiency scoring; Drug Repurposing and Combination Therapy adds an iPSC cortical-culture phenotypic screen.","why":"Gives each chapter a specific, source-grounded example of what current scoring and screening methods demonstrate in that therapeutic area.","chapters":[["Chemical Biology and Target Engagement","/therapeutics/chemical-biology.html"],["Cell and Gene Therapy AI","/therapeutics/cell-gene-therapy.html"],["Drug Repurposing and Combination Therapy","/therapeutics/repurposing-combinations.html"]],"sources":[{"label":"Chen et al., 2026 (VITAL)","url":"https://doi.org/10.1038/s42256-026-01291-z","type":"paper"},{"label":"Galli et al., 2026","url":"https://doi.org/10.1038/s41587-026-03237-7","type":"paper"},{"label":"Hsu et al., 2026","url":"https://doi.org/10.1038/s41587-026-03261-7","type":"paper"},{"label":"Greenberg et al., 2026","url":"https://doi.org/10.1038/s41467-026-76837-1","type":"paper"}]},{"category":"evidence","title":"Lab automation and literature-extraction bounds","change":"Self-Driving Labs adds GOLLuM, a calibrated Gaussian-process optimizer for next-experiment selection; Synthetic Biology Tools adds PUREdrop, a cell-free droplet screen of designed protein libraries; Knowledge Graphs and Literature AI adds PubMind, which extracts variant-function-disease associations from biomedical text.","why":"Gives three chapters a calibrated optimizer, a cell-free screening method, and a literature-extraction tool as concrete evidence bounds.","chapters":[["Self-Driving Laboratories","/automation/self-driving-labs.html"],["Synthetic Biology Design Tools","/automation/synbio-tools.html"],["Knowledge Graphs and Literature AI","/foundations/knowledge-and-literature.html"]],"sources":[{"label":"Ranković et al., 2026","url":"https://doi.org/10.1038/s42256-026-01283-z","type":"paper"},{"label":"Al Nahas et al., 2026","url":"https://doi.org/10.1038/s41467-026-76787-8","type":"paper"},{"label":"Wang et al., 2026","url":"https://doi.org/10.1038/s41467-026-76834-4","type":"paper"}]},{"category":"evidence","title":"Science sandboxes added as a closed-loop reasoning evaluation","change":"Benchmarks adds science sandboxes (MPRAbox, CodonBox) as a closed-loop evaluation of whether agents infer underlying rules or only raise an oracle score, noting the proposal is a preprint using damp and dry oracles.","why":"Gives the chapter an evaluation design aimed at distinguishing genuine rule inference from benchmark gaming, ahead of peer review.","chapters":[["Benchmarks for Bio AI","/governance/benchmarks.html"]],"sources":[{"label":"Rao et al., 2026, preprint","url":"https://arxiv.org/abs/2608.30165","type":"preprint"}]}]},
{"date":"2026-09-03","headline":"Behind-the-scenes maintenance","entries":[{"category":"maintenance","title":"Behind-the-scenes maintenance","change":"Behind-the-scenes maintenance; no change to chapter content.","why":"","chapters":[],"sources":[]}]},
{"date":"2026-09-04","headline":"NYT opinion: trial latency as an AI-era translation bottleneck","entries":[{"category":"developments","title":"NYT opinion: AI-era trial latency as a translation bottleneck","change":"Clinical Trial AI adds a section on a New York Times opinion essay arguing that patient trial latency, not discovery science, is often the bottleneck for new cancer therapies, reporting an asymmetric incentive against fast action and citing HHS's Operation TrialBlazer and an FDA proposal for expedited first-in-human IND review as an incremental policy response.","why":"Gives readers a researcher-sourced argument for why translation may be the binding constraint on AI-era cancer therapies, while noting interview-based opinion is not measured registry evidence and does not replace FDA and EMA validation discipline.","chapters":[["Clinical Trial AI for Translational Research","/therapeutics/clinical-trials.html"]],"sources":[{"label":"Teslo, 2026","url":"https://www.nytimes.com/2026/09/04/opinion/clinical-trials-drugs-science.html","type":"news"},{"label":"HHS, 2026","url":"https://www.hhs.gov/press-room/hhs-launches-clinical-trials-reform-initiative.html","type":"policy"},{"label":"FDA, 2026","url":"https://www.fda.gov/industry/fda-actions-support-and-strengthen-domestic-drug-manufacturing","type":"policy"}]}]},
{"date":"2026-09-05","headline":"Anthropic's Model Hardware Standard preview; evidence on invented citations in RAG tools","entries":[{"category":"evidence","title":"Evidence on invented citations added to literature RAG guidance","change":"Knowledge Graphs and Literature AI adds citations showing early ChatGPT evaluations found high rates of invented bibliographic references that looked real, and reports that retrieval-augmented clinical evidence tools can reduce the rate of invented citations relative to general chatbots, while noting citation integrity is not the same as clinical correctness.","why":"Grounds the chapter's resolve-before-cite guidance in measured rates of invented citations rather than a general warning about AI-generated citations.","chapters":[["Knowledge Graphs and Literature AI","/foundations/knowledge-and-literature.html"]],"sources":[{"label":"Gravel et al., 2023","url":"https://doi.org/10.1016/j.mcpdig.2023.05.004","type":"paper"},{"label":"Walters & Wilder, 2023","url":"https://doi.org/10.1038/s41598-023-41032-5","type":"paper"},{"label":"Artsi et al., 2026","url":"https://doi.org/10.1038/s41746-026-03077-4","type":"paper"}]},{"category":"developments","title":"Anthropic previews a Model Hardware Standard for lab instruments","change":"Anthropic's research preview of the Model Hardware Standard describes a model-agnostic driver that lets agents operate programmable instruments; partner notes report a liquid-handler and plate-reader workflow and an agent-supervised qPCR run. Self-Driving Labs treats it as an instrument-interface preview rather than a peer-reviewed self-driving lab, and a companion handbook covers instrument-access governance.","why":"Gives readers the developer's own framing that the spec is not open source and that safety evaluations are still being built, distinct from a demonstrated autonomous lab.","chapters":[["Self-Driving Laboratories","/automation/self-driving-labs.html"]],"sources":[{"label":"Anthropic, 2026","url":"https://www.anthropic.com/news/model-hardware-standard-research-preview","type":"announcement"}]}]},
{"date":"2026-09-06","headline":"Australia's Clinical Trial Notification scheme","entries":[{"category":"policy","title":"Australia's Clinical Trial Notification scheme added as a regulatory comparison","change":"Clinical Trial AI adds Australia's Clinical Trial Notification scheme as a comparison: the human research ethics committee assesses the trial while the Therapeutic Goods Administration does not evaluate trial data at submission, though ethics approval and institutional authorization remain required, and notes this does not establish a comparative reduction in trial-start time or equivalent safety outcomes.","why":"Completes the chapter's regulatory-process comparison with a primary source, while cautioning against reading a review-responsibility difference as a demonstrated speed or safety advantage.","chapters":[["Clinical Trial AI for Translational Research","/therapeutics/clinical-trials.html"]],"sources":[{"label":"TGA, 2024","url":"https://www.tga.gov.au/resources/guidance/australian-clinical-trial-handbook","type":"policy"}]},{"category":"accuracy","title":"Source verification","change":"Citations, figures, and links checked against their primary sources.","why":"","chapters":[],"sources":[]}]},
{"date":"2026-09-07","headline":"Behind-the-scenes maintenance","entries":[{"category":"maintenance","title":"Behind-the-scenes maintenance","change":"Behind-the-scenes maintenance; no change to chapter content.","why":"","chapters":[],"sources":[]}]},
{"date":"2026-09-08","headline":"Nature Biotechnology proteomic aging-clock analysis added for rentosertib","entries":[{"category":"evidence","title":"Proteomic aging clocks added to rentosertib evidence row","change":"Small Molecule Generation and ADMET adds a Nature Biotechnology analysis nesting six proteomic aging clocks in longitudinal serum from a Phase 2a subset of the TNIK inhibitor rentosertib, framing clock shifts toward younger predicted biological age as exploratory biomarker signals, not proof of clinical rejuvenation or a geroprotection claim.","why":"Keeps the rentosertib evidence row bounded to exploratory biomarker findings rather than a registration-level anti-aging claim.","chapters":[["Small Molecule Generation and ADMET","/therapeutics/small-molecules.html"]],"sources":[{"label":"Zhavoronkov et al., 2026","url":"https://doi.org/10.1038/s41587-026-03286-y","type":"paper"}]}]},
{"date":"2026-09-09","headline":"DeepMind releases AlphaGenome Atlas; OpenAI and DeepMind report on agentic research","entries":[{"category":"developments","title":"DeepMind releases AlphaGenome Atlas variant-effect catalogue","change":"Variant Effect Prediction adds DeepMind's AlphaGenome Atlas, a precomputed catalogue of predicted molecular effects for about 9 billion single-nucleotide changes with an AlphaGenome Variant Impact score for ranking, described as a research resource built on AlphaGenome predictions, not a licensed clinical diagnostic.","why":"Gives readers DeepMind's own framing of the atlas as a research resource, not a diagnostic tool, ahead of independent clinical validation.","chapters":[["Variant Effect Prediction","/molecular/variant-effects.html"]],"sources":[{"label":"DeepMind, 2026","url":"https://deepmind.google/blog/alphagenome-atlas-a-predictive-map-of-every-possible-dna-letter-change-in-the-human-genome/","type":"announcement"},{"label":"Nature News, 2026","url":"https://www.nature.com/articles/d41586-026-02835-4","type":"news"}]},{"category":"evidence","title":"GPN-Star phylogeny-aware variant-effect model added","change":"Variant Effect Prediction adds GPN-Star, a phylogeny-aware genomic language model using whole-genome alignments and species trees for coding and non-coding variant-effect prediction, treating its scores, like AlphaGenome and GPN-MSA, as ACMG/AMP triage evidence rather than standalone classifications.","why":"Adds a third phylogeny-aware scoring method to the chapter's triage-evidence framework for variant classification.","chapters":[["Variant Effect Prediction","/molecular/variant-effects.html"]],"sources":[{"label":"GPN-Star paper, Nature, 2026","url":"https://www.nature.com/articles/s41586-026-11005-5","type":"paper"}]},{"category":"evidence","title":"Preprint on the verification bottleneck for AI scientists","change":"Agentic Workflows adds a 2026 preprint arguing that AI-scientist progress is constrained as much by the verification budget as by hypothesis quality, and that evaluation should test experiment selection, revision under new evidence, and calibrated uncertainty across simulation and wet-lab verifiers.","why":"Gives the chapter a preprint perspective on evaluation design, explicitly not peer-reviewed and not evidence that any deployed agent closes the loop.","chapters":[["Agentic Science Workflows","/automation/agentic-workflows.html"]],"sources":[{"label":"Fang et al., 2026, preprint","url":"https://doi.org/10.20944/preprints202608.2107.v1","type":"preprint"}]},{"category":"developments","title":"OpenAI reports rising internal agent use in research","change":"Agentic Workflows adds OpenAI's September 2026 research-acceleration snapshot, which claims an automated-research-intern level inside its lab and rising agent-workday use under human priority-setting, labeled a vendor self-report with a pointer to a companion handbook for fuller safety context.","why":"Gives readers the developer's own claim about internal agent use, explicitly framed as a vendor self-report rather than independently verified capability.","chapters":[["Agentic Science Workflows","/automation/agentic-workflows.html"]],"sources":[{"label":"OpenAI, 2026","url":"https://openai.com/index/research-acceleration-view-inside-openai/","type":"announcement"}]},{"category":"evidence","title":"Preprint outlines DeepMind's AI Control Roadmap for internal agents","change":"Agentic Workflows adds DeepMind's GDM AI Control Roadmap (version 0.1), which proposes system-level detection and response ladders for imperfectly aligned internal agents, labeled a vendor roadmap with a pointer to a companion handbook for fuller control-related teaching.","why":"Gives readers the developer's own proposed safeguards for internal agent misalignment, framed as a vendor roadmap rather than an independently evaluated control method.","chapters":[["Agentic Science Workflows","/automation/agentic-workflows.html"]],"sources":[{"label":"Phuong et al., 2026, preprint","url":"https://arxiv.org/abs/2607.13087","type":"preprint"}]}]},
{"date":"2026-09-10","headline":"Self-Driving Labs cites the AI-scientist verification-budget preprint","entries":[{"category":"evidence","title":"Self-driving-labs chapter cites the AI-scientist verification-budget preprint","change":"Self-Driving Labs adds a pointer to the 2026 preprint arguing that closing the loop in biomedical discovery is constrained by the verification budget as much as by hypothesis quality, noting hard wet-lab verifiers are not interchangeable with soft simulators, with fuller teaching in Agentic Workflows.","why":"Extends the same verification-budget caution to autonomous-lab claims, since self-driving-lab and AI-scientist claims share the same evidentiary gap.","chapters":[["Self-Driving Laboratories","/automation/self-driving-labs.html"]],"sources":[{"label":"Fang et al., 2026, preprint","url":"https://doi.org/10.20944/preprints202608.2107.v1","type":"preprint"}]}]},
{"date":"2026-09-12","headline":"Behind-the-scenes maintenance","entries":[{"category":"maintenance","title":"Behind-the-scenes maintenance","change":"Behind-the-scenes maintenance; no change to chapter content.","why":"","chapters":[],"sources":[]}]},
{"date":"2026-09-13","headline":"JAMA+ AI Conversations interview on clinical-trial enrichment with machine learning","entries":[{"category":"evidence","title":"Perlis-Geraci JAMA+ AI Conversations on clinical-trial enrichment","change":"Clinical Trial AI for Translational Research adds Perlis's JAMA+ AI Conversations interview with Joseph Geraci on machine learning for clinical-trial enrichment, covering subpopulation identification and what regulators may expect when trials apply AI biomarkers; the chapter treats it as a teaching pointer, not methods validation of any vendor algorithm.","why":"Gives readers a plain-language framing of AI-assisted trial enrichment and how FDA reviewers may view it.","chapters":[["Clinical Trial AI for Translational Research","/therapeutics/clinical-trials.html"]],"sources":[{"label":"Perlis, 2026","url":"https://doi.org/10.1001/jama.2026.13579","type":"paper"}]}]},
{"date":"2026-09-15","headline":"Google DeepMind AI-in-science survey on productivity and verification bottlenecks","entries":[{"category":"developments","title":"Google DeepMind AI in Science survey on productivity and verification","change":"Agentic Workflows adds a September 2026 Google and Google DeepMind snapshot surveying 637 US and UK scientists alongside about 15 million Gemini interactions; scientists reported average time savings near 6.9 hours per week, but about 46% of those who saved time spent over a quarter of it verifying AI outputs. The chapter treats this as self-reported evidence, not causal proof.","why":"Flags the productivity data as self-reported and platform-specific, not a demonstrated rise in discovery rates.","chapters":[["Agentic Science Workflows","/automation/agentic-workflows.html"]],"sources":[{"label":"Codreanu, Imas, and Mateos-Garcia et al., Google DeepMind AI in Science, 2026","url":"https://ai.google/static/documents/AI-in-Science.pdf","type":"announcement"}]}]},
{"date":"2026-09-16","headline":"Paper2Agent turns papers into MCP agents; FDA CTD archive access essay added","entries":[{"category":"developments","title":"Institute for Progress essay on FDA CTD archive access","change":"Clinical Trial AI for Translational Research adds a 2025 Institute for Progress essay on FOIA Exemption 4 limits on Common Technical Document dossier access after Food Marketing Institute v. Argus Leader Media (2019), and its proposed AI Regulatory Transparency Fund to buy orphaned CTDs from bankrupt sponsors for a searchable corpus. The drafting-speedup example is vendor self-report, not independent evaluation.","why":"Surfaces a data-access barrier that keeps small biotechs from training regulatory AI tools on precedent filings.","chapters":[["Clinical Trial AI for Translational Research","/therapeutics/clinical-trials.html"]],"sources":[{"label":"Teslo, Institute for Progress, Biotech's Lost Archive, 2025","url":"https://ifp.org/biotechs-lost-archive/","type":"report"}]},{"category":"evidence","title":"Paper2Agent: converting research papers into MCP agent servers","change":"Agentic Workflows and the Model and Dataset Index add Paper2Agent, which converts a paper plus its codebase into a Model Context Protocol server a chat agent can query; case studies on AlphaGenome, Scanpy, and TISSUE reproduced tutorial results and answered novel queries. The chapter treats this as dissemination infrastructure, not evidence that agent teams raise wet-lab discovery rates.","why":"Gives researchers a way to query a paper's methods directly, while flagging that agentifying incomplete code fails.","chapters":[["Agentic Science Workflows","/automation/agentic-workflows.html"],["Model and Dataset Index","/appendices/model-dataset-index.html"]],"sources":[{"label":"Miao, Davis, Zhang, Pritchard, and Zou, 2026","url":"https://doi.org/10.1038/s41586-026-11044-y","type":"paper"},{"label":"Paper2Agent code repository","url":"https://github.com/jmiao24/Paper2Agent","type":"dataset"}]}]},
{"date":"2026-09-17","headline":"Cell perspectives on agentic science, virtual-cell world models, and aging-AI benchmarks","entries":[{"category":"evidence","title":"Guo et al. Cell perspective on agentic biomedical AI","change":"Agentic Workflows and Self-Driving Labs add a Cell perspective mapping biomedical AI's shift from tools that answer human-posed questions to agents that pose their own, alongside risks of hallucination, bias, dual use, and cognitive deskilling; the chapters note its conclusion centers responsible human-AI collaboration, not unsupervised wet-lab agency.","why":"Frames agentic-science risk in named terms while flagging that autonomy has not replaced scientific judgment.","chapters":[["Agentic Science Workflows","/automation/agentic-workflows.html"],["Self-Driving Laboratories","/automation/self-driving-labs.html"]],"sources":[{"label":"Guo, Ting, Su, Aliper, Zhavoronkov, and Ting, 2026","url":"https://doi.org/10.1016/j.cell.2026.08.052","type":"paper"}]},{"category":"evidence","title":"Virtual cells as biomedical world models","change":"Perturbation Prediction and Virtual Cells adds two Cell perspectives: Xing and Song frame the virtual cell as a multi-modal, multi-scale, stateful world model for simulation-first intervention design, and Noori, Zitnik, and colleagues define biomedical world models as action-conditioned simulators for counterfactual reasoning, distinct from static predictors. Evaluation Principles adds the same action-conditioned evaluation requirement.","why":"Sets an evidence bar so world-model claims are judged by named interventions, not by definitional language alone.","chapters":[["Perturbation Prediction and Virtual Cells","/cells/virtual-cells.html"],["Evaluation Principles for Life Sciences AI","/foundations/evaluation-principles.html"]],"sources":[{"label":"Xing and Song, 2026","url":"https://doi.org/10.1016/j.cell.2026.08.042","type":"paper"},{"label":"Noori, Fishman, Fang, Fesser, and Zitnik, 2026","url":"https://doi.org/10.1016/j.cell.2026.08.032","type":"paper"}]},{"category":"evidence","title":"LongevityBench for aging-AI readiness","change":"Benchmarks and Aging and Longevity Biology AI add LongevityBench, an open evaluation stack with Longevity-LLMs and Longevity Claw testing whether current AI systems are ready to lead aging research rather than only score aging clocks; the chapters treat strong results as readiness diagnostics, not lifespan, healthspan, or rejuvenation evidence.","why":"Adds a benchmark for aging-AI readiness while guarding against overreading scores as clinical outcomes.","chapters":[["Benchmarks for Bio AI","/governance/benchmarks.html"],["Aging and Longevity Biology AI","/organisms/aging-longevity.html"]],"sources":[{"label":"Zhavoronkov, Gladyshev, et al., 2026","url":"https://doi.org/10.1016/j.cell.2026.08.026","type":"paper"}]}]},
{"date":"2026-09-18","headline":"Virtual Biotech multi-agent system curates trial-outcome associations across 55,984 trials","entries":[{"category":"evidence","title":"Zhang et al. Science paper on Virtual Biotech multi-agent system","change":"Agentic Workflows and Clinical Trial AI for Translational Research add Virtual Biotech, a chief-scientific-officer agent coordinating domain scientist agents across genetics, omics, chemoinformatics, and clinical data; more than 37,000 clinical-trialist agents curated outcomes from 55,984 trials, finding drugs aimed at cell-type-specific, switch-like genes advanced with fewer adverse events. This is agent-curated association, not wet-lab proof of approval rates.","why":"Reports agent-curated trial associations as observational, not evidence that agent companies raise approval rates.","chapters":[["Agentic Science Workflows","/automation/agentic-workflows.html"],["Clinical Trial AI for Translational Research","/therapeutics/clinical-trials.html"]],"sources":[{"label":"Zhang et al., 2026","url":"https://doi.org/10.1126/science.aeg6779","type":"paper"}]}]},
{"date":"2026-09-21","headline":"AI-disclosure trends in math preprints; MMBU biomedical VLM perception benchmark","entries":[{"category":"evidence","title":"Epoch AI math-arXiv tracker on AI research-assistance disclosure","change":"Knowledge Graphs and Literature AI adds Epoch AI's math-arXiv tracker, which found research-assistance acknowledgments among established-author math preprints rose from about 1% in April 2026 to about 12% in August 2026, while any-purpose AI acknowledgments reached about 25% of all math preprints that month. The chapter frames this as voluntary disclosure in mathematics, not measured model use in biomedicine.","why":"Shows disclosure norms shifting even outside biomedicine, without implying similar rates apply to biomedical research.","chapters":[["Knowledge Graphs and Literature AI","/foundations/knowledge-and-literature.html"]],"sources":[{"label":"Epoch AI, math-arXiv research-assistance tracker, 2026","url":"https://epoch.ai/data/arxiv?view=graph&useCase=research_assistance&onlyEstablished=true","type":"report"},{"label":"Epoch AI, arXiv tracker methodology, 2026","url":"https://epoch.ai/data/arxiv-documentation","type":"report"}]},{"category":"evidence","title":"MMBU preprint on biomedical vision-language model perception","change":"Benchmarks adds MMBU, a preprint spanning 35 submodalities from 410 curated datasets, finding that medical fine-tuning gains on PathVQA, VQA-RAD, and SLAKE do not reliably survive a broader perception suite: closed-to-open gaps are large, object detection stays near or below random, and adapted and base models tie on most comparisons.","why":"Signals a benchmark-culture problem in biomedical imaging VLMs, not a wet-lab or clinical finding.","chapters":[["Benchmarks for Bio AI","/governance/benchmarks.html"]],"sources":[{"label":"D'Cunha et al., MMBU, 2026, preprint","url":"https://arxiv.org/abs/2606.06696v1","type":"preprint"}]}]},
{"date":"2026-09-22","headline":"Anthropic and OpenAI trusted-access programs; CRISPR-GPT and SAPP protein tools added","entries":[{"category":"revision","title":"Cross-chapter reference links added","change":"Agentic Workflows, Benchmarks, and Protein Design and Engineering add short cross-references pointing readers to the Toolkit, Histopathology AI, Microscopy and Cryo-EM AI, and Agentic Science Workflows sections that discuss related evidence.","why":"Helps readers find related coverage of the same tools and benchmarks across chapters.","chapters":[["Agentic Science Workflows","/automation/agentic-workflows.html"],["Benchmarks for Bio AI","/governance/benchmarks.html"],["Protein Design and Engineering","/molecular/protein-design.html"]],"sources":[]},{"category":"developments","title":"Anthropic Life Sciences Verification Program and Mythos 5.1 access","change":"Foundations and Toolkit add Anthropic's September 2026 Life Sciences Verification Program beta, which grants verified organizations Standard Use or High-risk Use access to Mythos, Opus, and Sonnet with biology classifiers more permissive than generally available models, alongside Claude Mythos 5.1 trusted access. This is a vendor access and monitoring program, not an independent safety certification.","why":"Distinguishes a vendor's own access-tier and monitoring program from independent validation of biological safety or performance.","chapters":[["AI for the Life Sciences","/foundations/ai-for-life-sciences.html"],["Toolkit for AI-Augmented Bio Research","/practical/toolkit.html"]],"sources":[{"label":"Anthropic, Life Sciences Verification Program, 2026","url":"https://www.anthropic.com/news/life-sciences-verification-program","type":"announcement"},{"label":"Anthropic, Claude Fable and Mythos 5.1 overview, 2026","url":"https://www.anthropic.com/claude-fable-and-mythos-5-1","type":"announcement"}]},{"category":"evidence","title":"CRISPR-GPT: LLM co-pilot for gene-editing experiment design","change":"Cell and Gene Therapy AI and Agentic Workflows add CRISPR-GPT, a 2025 Nature Biomedical Engineering LLM agent that plans and analyzes gene-editing experiments across knockout, base editing, prime editing, and CRISPRa/i; the authors report AI-guided wet-lab showcases including Cas12a knockout of four genes in a lung adenocarcinoma line and dCas9 activation of two genes in a melanoma line.","why":"Documents a peer-reviewed gene-editing co-pilot as a human-AI tool with defined cell-line demos, not unsupervised clinical editing.","chapters":[["Cell and Gene Therapy AI","/therapeutics/cell-gene-therapy.html"],["Agentic Science Workflows","/automation/agentic-workflows.html"]],"sources":[{"label":"Qu et al., 2025","url":"https://doi.org/10.1038/s41551-025-01463-z","type":"paper"},{"label":"CRISPR-GPT code release page (cong-lab)","url":"https://github.com/cong-lab/crispr-gpt-pub","type":"dataset"}]},{"category":"developments","title":"OpenAI GPT-Rosalind and Rosalind Workbench for life sciences","change":"Foundations and Toolkit add OpenAI's GPT-Rosalind trusted-access life-sciences model series, launched April 2026, and Rosalind Workbench, a ChatGPT research preview shipped August 2026 with guided scientific tasks, molecular and sequence viewers, and an NGS analysis path; Explore and Research modes gate advanced workflows to verified organizations. The chapters read these as vendor announcements, not peer-reviewed biology outcomes.","why":"Tracks a second frontier lab's trusted-access life-sciences program at parity with Claude Science, pending outcome evidence.","chapters":[["AI for the Life Sciences","/foundations/ai-for-life-sciences.html"],["Toolkit for AI-Augmented Bio Research","/practical/toolkit.html"]],"sources":[{"label":"OpenAI, introducing GPT-Rosalind, 16 April 2026","url":"https://openai.com/index/introducing-gpt-rosalind/","type":"announcement"},{"label":"OpenAI, new GPT-Rosalind capabilities, 3 June 2026","url":"https://openai.com/index/introducing-new-capabilities-to-gpt-rosalind/","type":"announcement"},{"label":"OpenAI Developers, Rosalind Workbench, 28 August 2026","url":"https://developers.openai.com/blog/rosalind-workbench","type":"announcement"}]},{"category":"evidence","title":"Saez-Rodriguez et al. Nature Methods perspective on foundation-model benchmarking","change":"Evaluation Principles and Benchmarks add a Nature Methods Perspective arguing that biomedical foundation models need epistemological limitation-testing beyond leaderboards: asking whether a model can be refuted, verified, or judged by utility, avoiding a self-assessment trap where developers grade their own finals, and importing CASP/DREAM-style independent referees with withheld evaluation data.","why":"Argues that leaderboard scores alone cannot settle whether a foundation model's outputs are trustworthy.","chapters":[["Evaluation Principles for Life Sciences AI","/foundations/evaluation-principles.html"],["Benchmarks for Bio AI","/governance/benchmarks.html"]],"sources":[{"label":"Saez-Rodriguez, Schäfer, Kalavros, and Stolovitzky, 2026","url":"https://doi.org/10.1038/s41592-026-03182-y","type":"paper"}]},{"category":"evidence","title":"SAPP and DMX: standardized protein production and characterization","change":"Protein Design and Engineering adds Semi-Automated Protein Production (SAPP), a modular expression, purification, and QC workflow that records yield, dispersity, and oligomeric state for hundreds of designs per day, plus DMX demultiplexing that cuts gene-synthesis cost roughly five-fold for thousand-design campaigns; reported showcases include redesigned fluorescent proteins and de novo binders that neutralize RSV in reported assays.","why":"Adds a standardized QC pipeline for protein design libraries, short of function-specific potency or safety evidence.","chapters":[["Protein Design and Engineering","/molecular/protein-design.html"]],"sources":[{"label":"Qian, Milles, Wicky, Ragotte, Baker, et al., 2026","url":"https://doi.org/10.1038/s41467-026-76740-9","type":"paper"},{"label":"Baker lab, DMX barcoding kit, Addgene","url":"https://www.addgene.org/kits/baker-dmx-barcoding/","type":"other"}]},{"category":"evidence","title":"MutexaGPT: LLM agents for enzyme-engineering simulation workflows","change":"Protein Design and Engineering adds MutexaGPT, an open multi-agent platform in Nature Computational Science that translates plain-language enzyme-engineering intuition into physics-based simulation workflows and mutation libraries; a showcase cavity-engineering library reported about a 40% hit rate with roughly four-fold enrichment of successful designs.","why":"Reports an LLM-agent design pipeline whose enrichment claim comes from one showcase campaign, not broad validation.","chapters":[["Protein Design and Engineering","/molecular/protein-design.html"]],"sources":[{"label":"Shao et al., 2026","url":"https://doi.org/10.1038/s43588-026-01049-y","type":"paper"},{"label":"Sztain, 2026","url":"https://doi.org/10.1038/s43588-026-01052-3","type":"paper"},{"label":"Wang et al., 2026","url":"https://doi.org/10.1038/s41467-026-75283-3","type":"paper"}]},{"category":"developments","title":"Claude Opus 5.5 system card: life-sciences capability block","change":"Benchmarks adds Opus 5.5 beside Mythos 5.1, given the same CB-1, not CB-2, classification with expanded bio safeguards, plus the system card's life-sciences capability block covering protein design, de novo binders, BioMysteryBench, and Protocols. The chapter treats this as a developer-reported classification and capability score, not an independent uplift trial.","why":"Flags a frontier lab's own bio-capability classification as self-reported, not third-party validated uplift evidence.","chapters":[["Benchmarks for Bio AI","/governance/benchmarks.html"]],"sources":[{"label":"Anthropic, Claude Opus 5.5 system card, 2026","url":"https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf","type":"announcement"}]}]},
{"date":"2026-09-23","headline":"Anthropic reports agentic discovery of a reverse-transcriptase-associated enzyme system","entries":[{"category":"developments","title":"Anthropic reports agentic genome-mining discovery of a novel enzyme system","change":"Agentic Workflows adds Anthropic's report that Claude Mythos 5 agents surveyed about 200,000 reverse-transcriptase clusters across roughly 1.9 billion protein clusters in about 21.5 hours without human interruption, flagging a tandem-repeat array beside a jumbo-phage reverse transcriptase now named array-associated reverse transcriptases (ART); function remains unknown, and wet-lab work stays with human scientists under BSL-1 or BSL-2.","why":"Reports this as company-published, not peer-reviewed, and not a basis for agents to handle biological materials without a human gate.","chapters":[["Agentic Science Workflows","/automation/agentic-workflows.html"]],"sources":[{"label":"Anthropic, Claude discovers novel enzyme system, 23 September 2026","url":"https://www.anthropic.com/news/claude-discovers-novel-enzyme-system","type":"announcement"},{"label":"Anthropic, ART technical preprint, 2026","url":"https://www-cdn.anthropic.com/22573675ada52a8ca8a97a1a4b4326b2f208a071.pdf","type":"announcement"}]}]},
{"date":"2026-09-24","headline":"He et al. PASCode scores single cells for association with clinical phenotypes","entries":[{"category":"evidence","title":"He et al. PASCode: phenotype-association scoring across single cells","change":"Single-Cell Foundation Models adds PASCode, a graph-attention framework scoring how strongly individual cells associate with clinical phenotypes in single-nucleus RNA-seq; on the PsychAD prefrontal cohort (more than 6 million nuclei from 1,494 donors) it prioritized microglia and astrocytes for Alzheimer disease association. The chapter treats it as a tool for mechanistic and atlas work, not a diagnostic or treatment model.","why":"Gives readers a released method, with code, pretrained models, and a web atlas, for ranking cells by phenotype association in population-scale single-cell data.","chapters":[["Single-Cell Foundation Models","/cells/single-cell-models.html"]],"sources":[{"label":"He et al., 2026","url":"https://doi.org/10.1038/s41591-025-04128-1","type":"paper"}]},{"category":"revision","title":"Plain wording for source review-status notes in Agentic Science Workflows","change":"Agentic Science Workflows states the review status of cited sources in plain terms, such as \"not a peer-reviewed article\", in passages that include Anthropic's reported enzyme-system discovery and Paper2Agent.","why":"Lets readers see each cited source's publication status without specialist shorthand.","chapters":[["Agentic Science Workflows","/automation/agentic-workflows.html"]],"sources":[]}]},
{"date":"2026-09-27","headline":"Source verification","entries":[{"category":"accuracy","title":"Source verification","change":"Citations, figures, and links checked against their primary sources.","why":"","chapters":[],"sources":[]}]},
{"date":"2026-09-28","headline":"AGENT thermostabilizes solid-state mRNA-LNPs; generative NRPS T-domain redesign","entries":[{"category":"evidence","title":"AGENT thermostabilizes solid-state mRNA-LNP formulations","change":"mRNA, RNA, and Vaccine Design describes AGENT: excipient screening plus Bayesian optimization for solid-state mRNA-LNP formulations. Six iterations in about one month yielded SM-102 and ALC-0315 compositions that kept full bioactivity after more than two months at 37 °C. Responses in rodents and non-human primates were non-inferior to fresh soluble vaccine. The evidence is particle storage, not sequence design.","why":"Separates a storage-stability result for formulated particles from sequence design and from an approved-product claim.","chapters":[["mRNA, RNA, and Vaccine Design","/therapeutics/vaccines.html"]],"sources":[{"label":"Tian et al., 2026","url":"https://doi.org/10.1038/s41587-026-03331-w","type":"paper"}]},{"category":"evidence","title":"Generative redesign of NRPS thiolation domains","change":"Protein Design and Engineering reports 76 de novo NRPS thiolation domains from ESM3, ProteinMPNN, and EvoDiff, and 578 variants tested in vivo. Designed domains supported product formation, kept hybrid junctions active, and raised yields by up to about three-fold versus the native domain. One design raised melting temperature by 12 °C. It is megasynthetase design, not autonomous invention of antibiotics.","why":"Shows a wet-lab gate on generative carrier-domain design, short of a claim that models invent antibiotics on their own.","chapters":[["Protein Design and Engineering","/molecular/protein-design.html"],["Synthetic Biology Design Tools","/automation/synbio-tools.html"],["Chemical Biology and Target Engagement","/therapeutics/chemical-biology.html"]],"sources":[{"label":"Bülbül et al., 2026","url":"https://doi.org/10.1038/s41467-026-77963-6","type":"paper"}]}]},
{"date":"2026-09-29","headline":"SpaCEy links spatial tissue graphs to survival; phage-host ML predicts E. coli infection","entries":[{"category":"evidence","title":"SpaCEy links spatial tissue graphs to progression and survival","change":"Spatial Omics and Tissue Models describes SpaCEy, an explainable graph model of tissue from molecular marker expression, without cell-type labels or anatomical regions as inputs. In a spatial proteomic lung-cancer cohort it surfaced progression-linked patterns, and across breast-cancer proteomic datasets it stratified survival within and across clinical subtypes. It is a research explainer, not a locked clinical prognostic device.","why":"Gives readers a spatial-clinical model whose outputs stay hypotheses until a locked prognostic evaluation exists.","chapters":[["Spatial Omics and Tissue Models","/cells/spatial-omics.html"],["Histopathology AI","/cells/histopathology.html"]],"sources":[{"label":"Rifaioglu et al., 2026","url":"https://doi.org/10.1038/s41467-026-77924-z","type":"paper"}]},{"category":"evidence","title":"Phage-host machine learning predicts strain-level E. coli infection","change":"Nucleic Acid and Genome Models describes phylogeny-agnostic machine learning that predicts strain-level phage-host infection from genomes. Experimental validation of 1,240 E. coli interactions reached AUROC 0.84, and model-guided cocktails of at most five phages covered up to 97.5% of held-out strains. The use is phage selection and cocktail design, not generative genome design, and cross-genus transfer remains limited.","why":"Adds a measured infection-prediction result for phage selection without treating it as generative genome design.","chapters":[["Nucleic Acid and Genome Models","/molecular/genome-models.html"]],"sources":[{"label":"Noonan et al., 2026","url":"https://doi.org/10.1038/s41564-026-02482-5","type":"paper"}]}]},
{"date":"2026-09-30","headline":"SynthIDBio watermarks designed proteins; Nobel Turing Challenge sets an autonomy bar","entries":[{"category":"evidence","title":"SynthID Bio watermarking for protein-design provenance","change":"Protein Design and Engineering describes SynthIDBio (Stutz et al., Nature, 2026) as a function-preserving watermark embedded in ProteinMPNN-style sampling: a proof of concept with redesign and resequencing limits, not a substitute for synthesis screening. DeepMind's Introducing SynthID Bio is the institutional companion: Swiss-cheese framing, DNA synthesis-provider triage, and an Evo 2 phage watermark watch item.","why":"Readers can check a peer-reviewed provenance mark for designed proteins and the lab's account of how that mark would sit beside DNA synthesis screening.","chapters":[["Protein Design and Engineering","/molecular/protein-design.html"]],"sources":[{"label":"Stutz et al., 2026","url":"https://doi.org/10.1038/s41586-026-10965-y","type":"paper"},{"label":"DeepMind, Introducing SynthID Bio, 2026","url":"https://deepmind.google/blog/introducing-synthid-bio/","type":"announcement"}]},{"category":"evidence","title":"Nobel Turing Challenge bar for autonomous scientific discovery","change":"Self-Driving Laboratories states the Nobel Turing Challenge (Kitano, 2021) as an autonomous Nobel-level discovery, horizon around 2050, and distinguishes 2024 Nobels for building AI systems from a future prize for a discovery made by AI. A 2025 Nature News Feature records expert timelines from a decade to fifty years: challenge vision, not evidence a deployed system has met the bar.","why":"Sets a public bar for autonomous discovery so readers do not treat prizes for building AI systems as prizes for discoveries made by AI.","chapters":[["Self-Driving Laboratories","/automation/self-driving-labs.html"]],"sources":[{"label":"Kitano, 2021","url":"https://doi.org/10.1038/s41540-021-00189-3","type":"paper"},{"label":"Ahart, Nature, 2025","url":"https://doi.org/10.1038/d41586-025-03223-0","type":"news"}]}]},
{"date":"2026-10-01","headline":"Calibrated perturbation metrics; genomics AI policies; wet-lab autonomy terms","entries":[{"category":"evidence","title":"Calibrated metrics for deep genetic-perturbation models","change":"Single-Cell Foundation Models states that low dynamic range in MSE and Pearson Δctrl can make deep perturbation models look weak against mean or PCA-linear baselines. Miller et al. (Nature Biotechnology, 2026) report that WMSE, weighted R²Δ, and normalized inverse rank let them beat those baselines on 14 datasets and 18 metrics. Ahlmann-Eltze et al. diagnose miscalibration.","why":"Stops a common metric from being read as proof that deep perturbation models add nothing, and stops calibrated gains from being read as universal superiority.","chapters":[["Single-Cell Foundation Models","/cells/single-cell-models.html"],["Perturbation Prediction and Virtual Cells","/cells/virtual-cells.html"],["Evaluation Principles for Life Sciences AI","/foundations/evaluation-principles.html"]],"sources":[{"label":"Miller et al., 2026","url":"https://doi.org/10.1038/s41587-026-03307-w","type":"paper"}]},{"category":"evidence","title":"AI-specific policies in national genomics initiatives","change":"Biological Data Infrastructure reports that population genomics programs feeding AI training need explicit AI governance, not only generic data-use terms. Hazel et al. (npj Digital Public Health, 2026) found AI-specific public policies in only three of 90 national genomics initiatives surveyed.","why":"Shows how rarely national genomics programs publish AI-specific rules for data that may train models.","chapters":[["Biological Data Infrastructure","/foundations/data-infrastructure.html"]],"sources":[{"label":"Hazel et al., 2026","url":"https://doi.org/10.1038/s44482-026-00043-5","type":"paper"}]},{"category":"developments","title":"Cloud access treated as orthogonal to wet-lab autonomy","change":"Self-Driving Laboratories treats cloud access as orthogonal to autonomy: a robot that carries out a preset protocol is automation, model-directed choice of the next step is autonomy, and remote facility access is cloud. Batalis's 30 Sep 2026 note is a short teaching companion to that distinction, an essay rather than a closed-loop methods paper.","why":"Keeps remote facility access from being counted as evidence that a laboratory chooses its own next experiment.","chapters":[["Self-Driving Laboratories","/automation/self-driving-labs.html"]],"sources":[{"label":"Batalis, 2026","url":"https://stephbatalis.substack.com/p/what-does-autonomous-actually-mean","type":"post"}]}]}
],"highlights":[{"date":"2026-09-30","title":"SynthID Bio provenance on protein design (Stutz et al., Nature, and DeepMind)","tier":"landmark","category":"evidence","chapter":["Protein Design and Engineering","/molecular/protein-design.html"],"sources":[{"label":"Stutz et al., 2026","url":"https://doi.org/10.1038/s41586-026-10965-y","type":"paper","short_label":"Nature paper"},{"label":"DeepMind, Introducing SynthID Bio, 2026","url":"https://deepmind.google/blog/introducing-synthid-bio/","type":"announcement","short_label":"DeepMind blog"}],"section":"/molecular/protein-design.html#sec-protein-design-mpnn","vendor":true},{"date":"2026-09-22","title":"Verified-access life-sciences model programs from Anthropic and OpenAI","tier":"major","category":"developments","chapter":["AI for the Life Sciences","/foundations/ai-for-life-sciences.html"],"sources":[{"label":"Anthropic, Life Sciences Verification Program, 2026","url":"https://www.anthropic.com/news/life-sciences-verification-program","type":"announcement","short_label":"Anthropic program"},{"label":"Anthropic, Claude Fable and Mythos 5.1 overview, 2026","url":"https://www.anthropic.com/claude-fable-and-mythos-5-1","type":"announcement","short_label":"Model overview"},{"label":"OpenAI, introducing GPT-Rosalind, 16 April 2026","url":"https://openai.com/index/introducing-gpt-rosalind/","type":"announcement","short_label":"GPT-Rosalind"},{"label":"OpenAI, new GPT-Rosalind capabilities, 3 June 2026","url":"https://openai.com/index/introducing-new-capabilities-to-gpt-rosalind/","type":"announcement","short_label":"Capability update"},{"label":"OpenAI Developers, Rosalind Workbench, 28 August 2026","url":"https://developers.openai.com/blog/rosalind-workbench","type":"announcement","short_label":"Workbench"}],"section":"/foundations/ai-for-life-sciences.html#sec-ai-for-life-sciences-introduction","vendor":true},{"date":"2026-09-18","title":"Zhang et al. (Science): a multi-agent Virtual Biotech curates outcomes from 55,984 trials","tier":"landmark","category":"evidence","chapter":["Agentic Science Workflows","/automation/agentic-workflows.html"],"sources":[{"label":"Zhang et al., 2026","url":"https://doi.org/10.1126/science.aeg6779","type":"paper","short_label":"Read paper"}],"section":"/automation/agentic-workflows.html#agentic-science-workflows-demonstrated"}],"months":{"2026-05":"The Life Sciences AI Handbook published its first edition, then reorganized within days from eight parts into six, adding chapters including Knowledge Graphs and Literature AI, Histopathology AI, Systems Biology and Multiscale Modeling, and Aging and Longevity Biology AI. A blinded NIST inter-tool analysis found baseline nucleic acid synthesis screening performance above 95 percent.","2026-06":"Foundation Models for Biology added OpenAI's Preparedness Framework, Anthropic's Responsible Scaling Policy, and Google DeepMind's Frontier Safety Framework, and Benchmarks added a dangerous-capability evaluation section covering WMDP, LAB-Bench, and uplift studies. Agentic Workflows and Self-Driving Labs cited Google Co-Scientist and FutureHouse Robin, and Clinical Trial AI for Translational Research added an FDA request for information on AI-enabled early-phase trials plus ProJenX and Unlearn's digital twin comparator in an ALS trial.","2026-07":"Foundations, Workforce and Compute, and Toolkit added Anthropic's beta launch of Claude Science, an AI workbench for scientific research. Information Hazards updated DURC and PEPP oversight coverage following Executive Order 14292, and Small Molecule Generation and ADMET added antimicrobial discovery evidence, including SyntheMol-RL, ApexGO, and CAMPER compounds validated in mouse infection models. Benchmarks organized biosecurity evaluation into five layers, adding GPT-5.6 and Claude Opus 5 classifications.","2026-08":"The handbook published a DOI and archival PDF edition and added a new Evaluation Workflows chapter with downloadable worksheets. Glossary cited the USG Policy for Stopping High-Risk Life Sciences Research, and Information Hazards updated its citations for the DURC and PEPP transition notices. Protein Design and Engineering added Anthropic's reported Claude-designed de novo protein binder campaign, and a cross-chapter citation audit aligned plant, agriculture, and ecology attributions with their sources.","2026-09":"Anthropic opened a Life Sciences Verification Program for Mythos 5.1, and OpenAI introduced GPT-Rosalind trusted-access tools. Zhang et al. (Science) described a multi-agent Virtual Biotech that curated outcomes from 55,984 trials. SynthIDBio (Stutz et al., Nature) embeds a function-preserving watermark in ProteinMPNN-style sampling, with DeepMind's SynthID Bio note as the institutional companion. Anthropic reported agentic discovery of a novel reverse-transcriptase system.","2026-10":"Miller et al. (Nature Biotechnology) report that calibrated metrics let deep perturbation models beat uninformative baselines on 14 datasets and 18 metrics, where MSE and Pearson Δctrl can mislead. Hazel et al. found AI-specific public policies in three of 90 national genomics initiatives. Self-Driving Laboratories treats cloud access as orthogonal to autonomy, with a 30 Sep 2026 note as a teaching companion."}}