BindingDB

The dataset is used for compound-protein binding affinity prediction, and it contains 218,615 compound-protein pairs.

BibTex: