CAPRI第32～37轮竞赛中蛋白质复合物结构的预测和评估

为了探究蛋白质复合物的结构与相互作用,建立了蛋白质的分子对接方法,从2014年起参加蛋白质复合物结构预测竞赛的预测和打分竞赛.该方法首先采用结构模建方法预测单体蛋白质分子的结构,通过快速傅里叶变换进行全空间构象搜索.然后,优化蛋白质复合物构象,并采用全原子统计势函数进行评价.总结第32~37轮CAPRI竞赛结果发现,该方法在T104、T105、T111和T118比赛中挑选出了L_(RMSD)小于2.0×10~(-10)m的近天然结构.通过对比其他国际优秀课题组的方法,分析预测和打分比赛中取得的成绩和不足之处,并为后续的研究提出了改进方案与建议.

Protein Complex Structure Prediction and Evaluation in CAPRI Rounds 32-37

To explore protein complex structures and interactions, a protein docking method was proposed,and the prediction and scoring evaluations of CAPRI from 2014 were conducted. Firstly, the structure modeling method was carried out to predict the protein monomer structure, and Fast Fourier Transformation (FFT) was applied to perform the global conformational searching. Then, the protein complex conformations were refined and evaluated by all-atom statistical potential function. The summarized results from Rounds 32-37 of CAPRI show that this docking method predicts the near-native structures with LRMSD<2.0×10 -10m in the experiments of T104,T105,T111 and T118. Compared with other international excellent groups, the advantage and disadvantages in the prediction and scoring experiments were analyzed. Finally, the possible improvements for protein structure prediction in the oncoming experiments were proposed.