SOFTDOCK: understanding of molecular recognition through a systematic docking study.

Molecular recognition and docking are essential to the biological functions of proteins. SOFTDOCK was one of the first molecular docking methods developed for protein-protein docking. Its ability to represent the molecular surface with different shapes and properties and to dock a variety of molecular complexes with certain conformational changes was demonstrated in a previous study. In the present work, we studied the effects of the docking parameters through statistical analysis. Seventy one typical binary complexes of different categories in PDB were also systematically docked for a test; 57 of them produced correct solutions with one set of docking parameters whereas the other 14 complexes required adjustment of the docking parameters, by decreasing the softness of the recognition and hence the background noise. We found that these 14 complexes had special structural features. Our results suggest that a variety of mechanisms may be involved in molecular recognition rather than the shape complementarity only, which is very helpful in developing more powerful methods for predicting molecular recognition.