蛋白质表面模块划分及其在结合位点预测中的应用

Division of Protein Surface Patches and Its Application in Protein Binding Site Prediction

蛋白质-蛋白质复合物的结合位点预测是计算分子生物学的一个难题.本文对蛋白质-蛋白质复合物数据集Benchmark3.0中的双链蛋白质复合物进行了研究,计算了单体的残基溶剂可接近表面积和残基间的接触面积,并据此提出了蛋白质表面模块划分方法.发现模块的溶剂可接近表面积与其内部接触面积的乘积(PSAIA)值能够提供结合位点的信息.在78个双链蛋白质复合物中,有74个体系其受体或配体上具有最大或次大PSAIA值的模块是界面模块.将该方法获得的结合位点信息应用在CAPRI竞赛Target39的复合物结构预测中取得了较好的结果.本文提出的基于模块的蛋白质结合位点预测方法不同于以残基为基础且仅考虑表面残基的传统预测方法,为蛋白质-蛋白质复合物结合位点预测提供了新思路.

Binding site prediction for protein-protein complexes is a challenging problem in the area of computational molecular biology. Using a set of double-chain complexes in Benchmark 3.0, we calculated the solvent accessible surface areas and inter-residue contact areas for each monomer and propose a division method of protein surface patches. We found that the products of the solvent accessible surface areas and internal contact areas of patches, the PSAIA values, could provide protein binding site information. In a dataset of 78 complexes, either receptors or ligands of 74 complexes had interface patches with the first or second greatest PSAIA values among all surface patches. A good docking result was achieved when the binding site information obtained with this method was applied in Target 39 of the CAPRI experiment. This patch-based protein binding site prediction method differs from traditional methods, which are based on single residue and consider only surface residues. This provides a new method for binding site prediction in protein-protein interactions.