Using Interaction Fingerprints as a Strategy of Protein-Ligand Interaction Patterns Recognition and Post-Docking Analysis

Analysis of ligand-protein structures can elucidate many biological mechanisms and guide molecular docking studies. Intermolecular interactions can be encoded in different ways. Binary vectors that represent that information can be called “interaction fingerprints,” and working with them is a good strategy to work with a large set of structures. We developed NEQUIM Contact System (NCS) software that generates and analyzes interaction fingerprints of protein-ligand complexes. We employed different encoding vectors available using NCS and we implemented a new strategy that includes sequence alignment and generation of models based on data sets of known structures. Post-docking analysis of virtual screening (VS) of targets PDE5, CDK2, HIVRT, and P38 showed a large increase of area under the curve (AUC) for an optimized protocol for each target. The best increase in AUC results for each target are P38 (116%), PDE5 (64%), HIVRT (40%), and CDK2 (4%).