Effective merging of 3D receptor structures, 3D ligand structures, and ligand binding affinity information provides information for machine learning models to predict the affinity of new chemical compounds.
3D structure data from rcsb.org, Alphafold and in-house developed homology models were used as receptor templates.
Ligands with affinity data were gathered from ChEMBL, bindingdb.org and pubchem.org. The 3D structures of the ligands were placed within their respective receptors via various in-house developed methods.
Our developed Forcefield was used to energy minimize the ligand-receptor structures to yield a standardized database of > 250 000 ligand receptor complexes.