中文摘要：

一般的蛋白质对接程序能够提供大量的待选构象，但其中仅含有少量的正确构象。现在对接的主要工作在于如何从这些大量构象中挑出正确构象。我们先前的研究工作证明蛋白质界面比非界面表面具有更高的能量。在这里，我们使用由Chen等人提出的一个用于检验、设计对接程序的蛋白质复合物标准库中的非抗原一抗体复合物，将侧链能量运用到对接中，并比较了侧链能量和残基配对倾向性、残基组成倾向性、残基保守性在对接中的表现。单独使用这四项的正确构象的平均百排分位排序分别为：38．6±19．6、26．3±20．8、22．7±16．6和37．8±26．1，但是对于个别蛋白，侧链能量的表现要优于其它的三个参数。我们将四个参数综合起来考虑，发展了一个新的打分函数，平均百排分位排序为22．2±7．8，并且提高了筛选效率。

英文摘要：

Protein- protein docking algorithms usually generate a lot of possible conformations, but a few of which are native - like structures. The major challenge now is how to distinguish the near- native structures from the non - native ones. Our previous work demonstrated that protein interface residues had a higher energy score than other surface residues. In this study, we used side chain energy to score docking decoys of 62 protein complexes and compared with the performance of residue pair potential score, residue interface propensity and residue conservation score. The average one - hundred - scaled ranks of the near native conformations were 38.6 ± 19.6,26.3 ± 20.8,22.7 ± 16.6 and 37.8 ± 26.1 respectively. Although the side chain energy doesn＇ t performt very well in the whole docking benchmark, it is quite successful in some cases. We also developed a new scoring function by combining the four terms and obtained the best prediction result（average rank 22.2±7.8）.