Toward full-cycle automation of closed-loop molecular discovery – a case study on JAK targets
Abstract
Bridging AI and self-driving laboratories, we report an integrated, closed-loop molecular discovery workflow with automation spanning all four stages of the Design-Make-Test-Analyse (DMTA) cycle, exemplified by the identification of novel JAK inhibitors (JAK1, JAK2, JAK3, and TYK2). We combined AI-driven molecular design and retrosynthesis with RoboRXN-assisted synthesis and automated in vitro screening on the Ulysses platform, while retaining expert supervision for selected synthesis, purification, and characterization steps. We performed two DMTA cycles and experimentally evaluated 36 compounds, of which 29 were synthesized within the workflow (seven were purchased). In the first round, the generative model, guided by an affinity predictor, identified structural analogs of established JAK inhibitors. The second iteration yielded molecules with statistically significantly improved pIC50 and ligand efficiency for all four targets, underscoring the effectiveness of the closed-loop workflow.