Correlation between performance of QM/MM docking and simple classification of binding sites.

Use of SiteMap for binding site classification and its connection to QM/MM (quantum mechanics/ molecular mechanics) docking performance were investigated. Using the hydrophilic/hydrophobic character values along with balance between them which SiteMap calculates, we sorted 455 cocrystal complexes available from protein data bank into three groups and tested how Glide, a conventional docking program, and QPLD, a QM/MM docking program, perform on them. QPLD showed improvements on all three groups over Glide but with varying degrees. Analysis of the results was carried out, and establishment of correlations between the classification of binding sites and QM/MM docking performance was attempted. It was found that QM/MM docking delivers the most improvements for primarily hydrophobic binding sites with substantial hydrophilic interactions. Based on our findings, we make suggestions for use of QM/MM docking and directions for further developments.