This J Med Chem paper (14 October 2026) asks a practical question: of all the analogues of a phenotypic hit, which should you make to get a high-quality chemical probe? The authors note that PDD feeds pipelines with first-in-class modalities, but target and mode-of-action deconvolution remains resource-intensive.
Using examples of successful PDD programs, some of which delivered approved first-in-class drugs, they derive three principles: (a) matched pairs of active/inactive stereoisomers; (b) ligand-topology-guided optimization; (c) exit vectors to speed up ADME work and target deconvolution, and so early in vivo proof of concept.
The text also puts PDD in context: at Novartis, phenotypic screening is now an integral part of hit finding, but an analysis of in-house programs over the past 10 years shows that fully target-agnostic screens represent only about 11% of programs; most are target/mechanism-informed (about 30%) or combine TDD and PDD (about 14%).
Takeaway: for a phenotypic hit, plan the inactive/active stereoisomer pair and an exit vector for deconvolution probes in the very first analogue set.
Reference: 10.1021/acs.jmedchem.6c02122, J Med Chem, 14 Oct 2026.