Insilico Medicine cuts drug discovery timeline to 12-18 months with generative AI platform Chemistry42, advancing ISM3412 cancer drug to Phase 1 trial

Traditional small-molecule drug discovery typically requires 2.5–4 years from project initiation to preclinical candidate nomination and demands synthesis and testing of thousands of molecules. Identifying selective inhibitors for targets like MAT2A in MTAP-deletion cancers—a genetic alteration present in NSCLC, pancreatic, and bladder cancers—is particularly difficult with conventional structure-based methods.

Insilico Medicine applied its proprietary ligand-based generative AI chemistry platform, Chemistry42, to design ISM3412, a highly selective and orally bioavailable MAT2A inhibitor with a novel structure. The platform guided candidate nomination in May 2022, and the compound exploits synthetic lethality in MTAP-deficient tumor cells while sparing healthy tissue. ISM3412 received IND clearance from both the U.S. FDA and China's NMPA in April–May 2024, enabling a global multicenter Phase 1 trial.

ISM3412 progressed from AI-guided preclinical candidate nomination to first-patient dosing in a Phase 1 trial (NCT06414460) at Cancer Hospital Chinese Academy of Medical Sciences. Across 22 drug candidates nominated between 2021 and 2024, Insilico's AI-driven process averaged only 12–18 months from project initiation to PCC nomination—roughly 3× faster than the industry standard—requiring synthesis and testing of only 60–200 molecules per project. The success rate from preclinical candidate to IND-enabling stage reached 100%.

• Generative chemistry AI can compress the preclinical discovery phase from years to months while dramatically reducing the number of molecules that need to be synthesized and tested.
• Targeting synthetic lethality mechanisms (MTAP deletion / MAT2A inhibition) via AI-enabled ligand-based design demonstrates that AI can identify and exploit complex genomic vulnerabilities for novel oncology drug candidates.
• A 100% PCC-to-IND success rate across 22 programs suggests that AI-guided candidate selection materially de-risks the transition from discovery to formal development.