DianShi-RxnDB Automates Extraction of 24 Million Organic Reactions
September 5, 2026
DianShi-RxnDB is a large-scale organic reaction platform built from USPTO and EPO patents published between 1976 and 2025. The automated pipeline extracted 24 million instances, with 14.8 million passing quality checks for use in AI4Chem applications.
HOW THIS AFFECTS YOU
●
builderYou can use this high-density dataset to train specialized agents for organic synthesis and chemical planning.
●
healthThis advances automated drug discovery by providing fine-grained reaction data including yields and temperatures.