Re-Docking Scheme to Explore Docking Search Space by using Interaction Profiles

Abstract

In protein-protein interaction predictions, there are various approaches to obtain near-native 3D structure of protein complexes. One of the most available methods is rigid-body docking process, generating many protein complexes (decoys) as candidates of the native complex.However, we sometime faced with one of the critical problems to solve, which is a situation of no near-native decoys including a decoy dataset. Even if the bound-state case, in 9 out of 44 protein pairs, we could not obtain near-native decoys. To overcome this situation, we applied interaction fingerprint (IFP) to this problem. IFP method in docking process is originally developed for cluster analysis by comparing among decoys in our previous work [Uchikoga & Hirokawa, (2010) BMC Bioinfor. 11:264]. This method can applied to proteins with large conformation changes, for example, calmodulin. IFP composed of frequencies of interaction between amino acid residues. Therefore, much more different structures can compare each other.The critical situation of no near-native decoys results from a fact that docking search space is not large enough to obtain near-native decoys. Therefore, we proposed re-docking scheme for exploring docking search spaces by restricting protein surfaces after assembling interaction surfaces of decoys using IFPs. We applied re-docking scheme to several docking cases and will discuss the results.