Appendix C — Case Studies

Published

July 7, 2026

Short applied cases for interpreting life sciences AI claims. Each case maps a published result or capability claim to the three-tier evidence framework (Demonstrated, Theoretical, Beyond Current Capabilities) and the validation discipline the handbook applies elsewhere in detail. Cross-references to chapters in parentheses.

Case 1: Structure Prediction as a Research Input

A team without an experimental structure uses AlphaFold DB to generate hypotheses about a protein domain. The correct use is hypothesis generation, followed by confidence review (pLDDT distribution, PAE matrix), domain inspection, and assay planning. The high-pLDDT regions are usable for docking and mutagenesis design; the low-pLDDT regions are predictions of disorder and should be treated as such.

The failure mode this case avoids: treating the structure visualization as if it were a measured structure. AlphaFold confidence metrics are decision aids; reading the visualization without reading the confidence is misuse.

References: AlphaFold Protein Structure Database, 2026; Jumper et al., 2021. See Protein Structure Prediction chapter.

Case 2: Protein Binder Design

A protein engineering group uses RFdiffusion for backbone proposals and ProteinMPNN for sequence design. The design is not a success until expression, binding, specificity, and stability data support it. In-silico filtering (per-residue pLDDT distribution, predicted developability, basic biophysical filters) reduces the candidate set before bench testing.

Evidence tier placement: backbone generation and inverse-folding sequence design are Demonstrated. Function on a specific target is Theoretical until the assay confirms it.

References: Watson et al., 2023; Dauparas et al., 2022; Dauparas et al., 2025. See Protein Design and Engineering chapter.

Case 3: Docking Benchmark Failure

A docking method looks strong by RMSD but generates physically implausible poses. PoseBusters shows why chemical validity checks belong beside geometric metrics: bond lengths, stereochemistry, severe clashes, ring geometry. Many low-RMSD deep-learning docking poses fail at least one basic validity check.

The lesson generalises: a single metric can hide invalid outputs in many AI domains. Validity filtering is part of the evaluation, not an optional add-on.

References: Buttenschoen et al., 2024. See Small Molecule Generation and ADMET, and Translational Evidence and Failure Modes chapters.

Case 4: Single-Cell Perturbation Forecasting

A group uses GEARS to rank perturbations for a follow-up experiment. The result supports prioritisation, not general causal certainty. The Ahlmann-Eltze 2025 Nature Methods critique shows that for many perturbation-prediction tasks, PCA plus linear regression matches or beats deep methods. The right discipline is always to run the linear baseline first.

Evidence tier placement: GEARS-style perturbation prediction is Demonstrated for bounded settings (in-distribution gene pairs); generalisation to novel-context settings is Theoretical or weaker.

References: Roohani et al., 2024; Ahlmann-Eltze et al., 2025. See Perturbation Prediction and Virtual Cells chapter.

Case 5: Closed-Loop Experimentation in Chemistry

A self-driving lab chooses the next experiment based on prior measurements. Coscientist uses GPT-4 as planner on cloud robotic platforms and reproduces Pd-catalysed cross-coupling chemistry. The system needs protocol versioning, instrument logs, objective functions, and stopping rules. Counting loop iterations is the right discipline: a platform that runs once is automation, not autonomy.

Evidence tier placement: closed-loop optimisation for bounded chemistry tasks is Demonstrated. Open-ended autonomous biology beyond bounded settings is Beyond Current Capabilities.

References: Boiko et al., 2023; Burger et al., 2020. See Self-Driving Laboratories chapter.

Case 6: Multi-Agent Biology Research

Virtual Lab (Stanford) coordinates a PI agent and specialist agents to design 92 SARS-CoV-2 nanobodies. Two designs showed improved binding to JN.1 and KP.3 variants while retaining ancestral spike binding in experimental validation. The contribution is the multi-agent workflow with experimental followthrough, not the absolute binding numbers.

Evidence tier placement: multi-agent biology research with wet-lab validation for one target is Demonstrated. Generalisation across biology classes is Theoretical.

References: Swanson et al., 2025. See Agentic Science Workflows chapter.

Case 7: A-Lab and the Novelty-Claim Discipline

A-Lab reported 41 new inorganic materials autonomously produced in 17 days. A PRX Energy critique from solid-state chemists argued that the novelty claim did not hold because many entries were misclassified or were known phases. The episode is a clear case study of how autonomous-discovery novelty claims can outrun validation: throughput is not discovery.

Lesson: automated characterisation can be wrong in characteristic ways; independent expert review is part of validation; novelty claims should pass scrutiny from the relevant subdiscipline before being treated as discoveries.

References: Szymanski et al., 2023; Leeman et al., 2024. See Self-Driving Laboratories chapter.

Case 8: AI-Discovered Drug in Phase IIa

Insilico Medicine’s ISM001-055 (rentosertib) is an AI-discovered TNIK inhibitor for idiopathic pulmonary fibrosis. The peer-reviewed Nature Biotechnology paper covers the discovery and Phase I narrative. Positive Phase IIa topline was reported in November 2024 by the company. Phase IIa is a meaningful signal that the compound has measurable activity in the indication; it is not registration-grade and not yet peer-reviewed at Phase IIa quality.

Evidence tier placement: an AI-discovered candidate reaching Phase IIa is Demonstrated for this specific development path. Clinical success at registration scale is Theoretical for this specific candidate and Beyond Current Capabilities as a general AI-discovered-drug claim.

References: Ren et al., 2025; Insilico Medicine, November 2024. See Small Molecule Generation and ADMET, Translational Evidence and Failure Modes chapters.

Case 9: Variant Interpretation Inside the ACMG Framework

A clinical laboratory adds AlphaMissense outputs to its variant-interpretation pipeline. The correct use is as PP3 (computational evidence of pathogenicity) or BP4 (computational evidence of benignity) input alongside population data, functional studies, and segregation. The score is not a classification; the framework is.

Lesson: variant predictors are evidence inputs into the ACMG/AMP framework, not substitutes for it. Per-laboratory validation of score thresholds is mandatory before clinical use.

References: Cheng et al., 2023; Richards et al., 2015. See Variant Effect Prediction chapter.

Case 10: Single-Cell Foundation Model in Practice

A research group uses scGPT for cell-type annotation and considers extending it to perturbation prediction. For annotation, the model is Demonstrated. For perturbation, the published evidence is contested by Ahlmann-Eltze 2025: deep methods do not consistently beat PCA plus linear regression. The right discipline is to use scGPT for what it is demonstrated to do (representation, integration, atlas search) and run a linear baseline before adopting it for perturbation prediction.

References: Cui et al., 2024; Ahlmann-Eltze et al., 2025. See Single-Cell Foundation Models chapter.

Case 11: AlphaFold 3 Restricted Release and Open-Source Response

AlphaFold 3 was published in Nature in May 2024 but released as a server with rate limits and no training code. The community responded with open-source AF3-class models: Boltz-1 (MIT, November 2024) and Chai-1 (October 2024). For research programs that need on-premise inference, batch screening at scale, or commercial-use rights, the open-source alternatives are the practical implementations.

Lesson: licensing and code availability are first-class evaluation criteria, not afterthoughts. A model that cannot run in your environment is not usable for your program.

References: Abramson et al., 2024; Wohlwend et al., 2024, preprint; Chai Discovery et al., 2024, preprint. See Protein Structure Prediction, Reproducibility and Open Science chapters.

Case 12: Agentic Research Planning Without Lab Authorisation

A research agent drafts hypotheses and suggests protocols. The output requires source verification, human review, and authorisation before any biological action. Permissions should be separated: read, analysis, procurement, and laboratory execution are distinct authorisations. An agent with read permission alone is safe; an agent with laboratory-execution permission and no human gate is not.

Lesson: agentic systems are useful as research tools when tasks are bounded, sources are checked, and lab actions require explicit human authorisation. Treating fluent agent output as validated science is the failure mode.

References: M. Bran et al., 2024; ARPA-H IGoR, 2026. See Agentic Science Workflows, Information Hazards in Capability Research chapters.