TOUGH-M1

 

A non-redundant and representative dataset of ligand-binding pockets extracted from proteins with globally unrelated sequences and structures. The dataset comprises 7,524 experimental structures of protein-ligand complexes with pockets predicted by Fpocket. It is divided into two subsets, 505,116 pairs of pockets binding chemically similar molecules and 556,810 pairs of pockets binding different ligands.

 
TOUGH-M1