My notes on agentic AI, drug discovery, and teaching
-
Likelihood of approval, phase transitions, and the crowded-vs-abandoned map of therapeutic areas
Phase I likelihood of approval is 6.7% industry-wide. Oncology, rare disease, and biomarker-stratified programs tell different stories. A 2026 map.
-
Tissue specificity as a safety filter: GTEx, Human Protein Atlas, and scRNA-seq for target prioritization
Tissue specificity predicts on-target, off-tissue side effects before a drug is ever dosed. A practitioner's guide to GTEx, Protein Atlas, and scRNA-seq.
-
Druggability, ligandability, and modality choice in the AlphaFold 3 era
Druggability assessment used to ask "does this protein have a pocket?" In 2026 it asks "which of six modalities fits this target best?" A field guide.
-
How to tell a drug target matters: evidence frameworks for target–disease linkage
Target-disease association evidence is what Phase II efficacy hinges on. A practitioner's guide to 5R, GOT-IT, the omics stack, and knowledge graphs.
-
Drug target discovery: the front-of-funnel decision behind most Phase II failures
Drug target discovery is the least quantified stage of the drug pipeline and a major determinant of Phase II outcomes. Why it matters, and where ML fits.
