Design of a highly selective, allosteric, picomolar TYK2 inhibitor using novel FEP+ strategies

JAK/TYK kinases drive inflammatory diseases like psoriasis, but existing JAK inhibitors are non-selective across the nearly identical TYK2/JAK1/JAK2/JAK3 binding sites, causing safety issues like heart problems and thrombosis. Nimbus Therapeutics and Schrodinger used large-scale FEP+ calculations to predict potency and off-target effects, developed a novel absolute FEP+ approach for aqueous solubility, screened 4,000 candidates through MM-GBSA filters then scored 46 with FEP+, pivoted to the allosteric JH2 domain after external validation data, and virtually screened 2.4M compounds via Glide/WScore, using LiveDesign to manage the process. This identified 9 molecules meeting potency/selectivity/solubility criteria and picomolar JH2 scaffolds within 3 months of the pivot, leading to NDI-034858 (now TAK-279) with an "exquisitely clean selectivity profile"; it achieved positive Phase 2b results in November 2022 and Takeda acquired the program in February 2023.

The customer is anonymised in Schrödinger’s own case study — not withheld by us.

Schrödinger is a physics-plus-machine-learning computational platform for molecular design; all top-20 pharma companies by revenue are customers of its drug-discovery software.

This sits in the Build stage of the AI use-case lifecycle: Named examples from the foundation-model labs and platforms pharma teams use to actually construct, train and generate the model or application.