---
title: "The Life Sciences AI Handbook"
source: "https://lifesciencesaihandbook.com/"
author: "Bryan Tegomoh, MD, MPH"
last_updated: "2026-11-16"
doi: "10.5281/zenodo.22073135"
license: CC BY 4.0
generated-by: handbook-markdown-export
---

# The Life Sciences AI Handbook

Evidence for AI in Biology and Drug Discovery

What works, what remains unproven, and what stands in the way.

Author

[Bryan Tegomoh, MD, MPH](https://bryantegomoh.com/)

Updated

November 2026 · [Changelog](./changelog.html)

[Free PDF](./life-sciences-ai-handbook.pdf) · [Print and Kindle editions](https://www.amazon.com/dp/B0HQ1JGZRT)

![](./images/bryan-tegomoh-author.jpg)

[Bryan Tegomoh, MD, MPH](https://bryantegomoh.com/)Physician-scientist and epidemiologist in health AI and biosecurity.

[Follow @BryanTegomoh](https://x.com/BryanTegomoh)

## Welcome to The Life Sciences AI Handbook

Foundation models now predict protein structures before experiments solve them, represent cells in latent space, and generate molecules faster than chemistry can synthesize them. The bottleneck has moved from the model to functional biology.

The hard part is translation: separating what a model suggests from what biology has shown, and deciding what still has to be tested.

**Each chapter answers two kinds of questions:** what the field is trying to solve and where it stands, and what is demonstrated, what remains theoretical, and what to do next.

**This handbook is continuously updated as new papers, benchmarks, and model releases change what the evidence supports.**

Important Disclaimers

**This handbook is for research and educational purposes only.** It does not constitute clinical advice, regulatory guidance, or a recommendation to use any specific tool or platform. Model outputs discussed here are hypotheses that require experimental validation; they are not endpoints.

**Validation and safety remain context-specific.** Researchers and program leaders should design appropriate validation experiments, assess reproducibility and safety, meet applicable biosafety and biosecurity requirements, and follow publication and data-sharing norms in their field.

**Information may become outdated** given the pace of model releases, benchmark updates, and laboratory automation tooling. Verify model versions, training data scope, and benchmark results against primary sources before relying on them.

**For dual-use considerations,** see [Information Hazards in Capability Research](governance/information-hazards.html) and the companion [Biosecurity Handbook](https://biosecurityhandbook.com/).

------------------------------------------------------------------------

## Recent highlights

- 13 Nov [Federal DNA synthesis screening becomes a purchasing condition for funded labs (opens the source in a new tab)](https://www.whitehouse.gov/wp-content/uploads/2026/11/NSTM-6-NASS-Policy.pdf "White House, NSTM-6, 2026") Order screening moves from voluntary practice to a funding condition, so synbio teams need adherent suppliers before the 2027 deadlines. In [Synthetic Biology Design Tools](./automation/synbio-tools.html) Policy & regulation [White House, NSTM-6, 2026](https://www.whitehouse.gov/wp-content/uploads/2026/11/NSTM-6-NASS-Policy.pdf)
- 13 Nov [Anthropic guidance on building safety evaluations: stress-testing the grader (opens the source in a new tab)](https://www.anthropic.com/research/building-strong-safety-evaluations "Anthropic, Building strong safety evaluations, 2026") If a conclusion changes when the grader changes, the evaluation measured the grader rather than the research agent. In [Agentic Science Workflows](./automation/agentic-workflows.html) News & developments [Anthropic, Building strong safety evaluations, 2026](https://www.anthropic.com/research/building-strong-safety-evaluations)
- 6 Nov [RFdiffusion3 published in Science (opens the source in a new tab)](https://doi.org/10.1126/science.aeh2982 "Butcher et al., 2026, Science") The all-atom design claim now rests on a peer-reviewed paper rather than a developer preprint. In [Protein Design and Engineering](./molecular/protein-design.html#sec-protein-design-rfdiffusion) New evidence [Butcher et al., 2026, Science](https://doi.org/10.1126/science.aeh2982)
- 6 Nov [OpenAI’s BioBasicsBench: AI agents reading routine wet-lab data (opens the source in a new tab)](https://openai.com/index/measuring-how-ai-agents-handle-wet-lab-science/ "OpenAI, BioBasicsBench, 2026") Agents are starting to read raw lab data, and OpenAI’s own failure cases show assumptions overriding controls, so their outputs still need a scientist’s check. In [Benchmarks for Bio AI](./governance/benchmarks.html) News & developments [OpenAI, BioBasicsBench, 2026](https://openai.com/index/measuring-how-ai-agents-handle-wet-lab-science/)
- 28 Oct [Anthropic and five partners report agent-guided preclinical design results (opens the source in a new tab)](https://www.anthropic.com/research/early-progress-drug-development "Anthropic, Early progress in drug development, 2026") These are partner-reported results, not peer-reviewed studies, and they mark how far language-model agents reach into bounded preclinical design today. In [Agentic Science Workflows](./automation/agentic-workflows.html) News & developments [Anthropic, Early progress in drug development, 2026](https://www.anthropic.com/research/early-progress-drug-development)

[View the full Changelog →](./changelog.html)

## Start Here

**Read the evidence framework first, then the best-established molecular example.**

1.  [Executive Summary](executive-summary.html): Evidence framework and core conclusions across the handbook
2.  [AI for the Life Sciences](foundations/ai-for-life-sciences.html): Scope, audience, and how the chapters are organized
3.  [Protein Structure Prediction](molecular/protein-structure.html): The best-established capability, and what it does and does not solve

Then continue to [Evaluation Principles](foundations/evaluation-principles.html) before adopting any model in a research program.

For reading paths by role and links to each research area, see the full [Start Here page](./start.html); for teaching, see the [Course Syllabus Template](./appendices/course-syllabus.html) and [Case Studies](./appendices/case-studies.html).

An AI assistant helping you should follow the [Guidance for AI Agents](./for-ai-agents.html).

------------------------------------------------------------------------

## Companion Handbooks

### [The Physician AI Handbook](https://physicianaihandbook.com/)

Peer-Reviewed Evidence Across Every Medical Specialty

What works. What doesn’t. What happens when AI is wrong.

### [The Public Health AI Handbook](https://publichealthaihandbook.com/)

Evaluating AI Tools for Public Health Practice

Some AI tools work in the field. Others only work on paper.

### [The Biosecurity Handbook](https://biosecurityhandbook.com/)

Biological Security in the AI Era

A guide for two communities that don’t speak the same language.

Free to read and reuse under CC BY 4.0 · DOI 10.5281/zenodo.22073135 · [How to cite](./how-to-cite.html)

This handbook is open and continuously maintained. [Support the Life Sciences AI Handbook →](./support.html)
