Analysis of Physico-Chemical Properties of Protein Docking Decoys Generated by Rigid-Body Docking

Abstract

We have developed a rigid-body docking algorithm as MEGADOCK, which is a high speed docking simulation software. Rigid-body docking generates many possible protein complexes, referred as docking decoys. Because sets of docking decoys include false positives, near-native decoys are searched among these decoys.However, in some cases, near-native decoys are no found in a set of docking decoys. We then have developed a fingerprint method for searching near-native docking decoys by proposed a novel method, Re-docking scheme of exploring docking spaces efficiently. This scheme is useful for prediction protein-protein interactions in the case of no near-native complexes after a docking process. And this scheme of an iterative docking process also gives us more information of interaction surfaces.To understand a mechanism of protein-protein interactions, we analyzed sets of docking decoys generated by rigid-body docking algorithm. Interaction fingerprints is available for analysis of interactions with amino acids. We could obtain which amino acid pairs are appeared in each docking decoys. Then, we focused on some physico-chemical properties of interaction amino acid pairs, electrostatic and hydrophobicity, of a set of docking decoys. Each docking decoys includes information of sets of possible interacting amino acid pairs. We used MEGADOCK for docking process and analyzed interaction properties of decoys. Typical 44 protein pairs are used for datasets. In this work, we investigated mainly differences between true and false protein pairs.