一种蛋白质点突变计算机预测的并行模型

认识和预测蛋白质天然构象的波动对蛋白质-蛋白质对接和设计等应用是非常重要的.但是许多骨架柔性的方法会导致骨架较大幅度的波动.Backrub模型能够对骨架进行微小的扰动,符合高分辨率晶体结构中观察到的构象的微妙变化.本文提出了一种基于Backrub的并行扰动骨架和侧链的模型,可以对天然构象的等价状态进行模拟.这种并行扰动方式更加接近于真实情况下蛋白质构象的运动方式,更好地模拟了实验数据.通过预测10个点突变实例,相比串行随机扰动模型产生的构象,并行模型不仅从时间上提高了产生构象的速度,更提高了侧链的预测精度.

Parallel Model for Computational Prediction of Protein Point-mutation

Understanding and predicting conformational fluctuations of the native protein is critical for protein-protein docking and design.Though numerous methods have been developed to model such backbone flexibility,they may result in highly non-local displacements of the protein backbone.Backrub model can capture the subtle shifts observed in high-resolution crystal structures.Based on Backrub model,we present a new model of simultaneously perturbing protein backbone and side-chain,to simulate the conformational fluctuations of the native proteins.These simultaneous perturbations in principle closely mimic the real motion of the native conformations and satisfy the nature evolution procedure.We evaluate our model by predicting ten point-mutation cases.Compared to the sequential movement model,our parallel model improves the performance in both the predicting time and the accuracy of the protein side-chain.