蛋白质结构建模的后期优化策略

以Lip8为研究对象, 以分级开放式优化和组合优化为优化策略, 分别采取四种不同的优化手段对相同的初始模型进行了优化, 同时比较了含水体系与不含水体系的优化过程, 最终获得8个相应优化后的蛋白模型. 采用Procheck, Errat, Profile 3D及Ramachandran 图等方法对上述8个模型进行评估. 通过比较各项评估结果, 研究了不同优化方法对最终模型准确性的影响, 结果表明, 优化过程中采用分级开放式策略, 同时考虑显性溶剂(水)体系和加入分子动力学优化的优化方法可以获得相对最优结构.

Optimizing Strategies Research on the Later Stage of Protein Structure Model

Homology modeling is the most widely used and fully developed method for protein structure prediction.Researchers of this field paid great attention to either improvement of modeling method or update of modeling algorithm in the past while ignoring the fact that different structure optimizing methods could contribute to the accuracy of the final structure as well.Lip8 was chosen as an example to find out the influences of different optimizing methods.After a initial Lip8 model was built by the method of homology modeling,four ways were applied to optimize the model based on two different strategies called "grade unpacking" and "fitting together".Each of these ways was considered under two different conditions: with periodic bounded cubic(PBC) water surrounded or without,which resulted in 8 optimized structures all together.Procheck,Errat,Profile 3D and Ramachandran plot were used to evaluate each final model.From the results,we found that "grade unpacking" strategy combining with molecular dynamics optimization under explicit water could get a better model.