基于残基特异性力场的蛋白质结构从头预测

The prediction of a protein’s three-dimensional structure from its amino acid sequence is a great challenge in computational chemistry and computational biophysics. With the increasing performance of computers, development of protein force fields and enhanced sampling methods, recently it is possible to fold some fast-folding small proteins to near experimental structures through molecular dynamics (MD) simulations. However, this ab initio folding method is still far from practical for a useful protein structure prediction method. In this project, we will further develop implicit-solvent force field and enhanced sampling methods, based on the explicit-solvent residue-specific force fields RSFF1 and RSFF2 developed in this applicant’s previous NSFC project, to achieve ab initio folding of some proteins of slower-folding or relatively large sizes. To greatly accelerate the folding rate, some additional interactions will be added in the force field to reduce the folding free energy barrier. This project not only can promote the development of better methods for protein structure prediction, but also can provide new insight into the important scientific problem of protein folding mechanism.

由蛋白质的氨基酸序列预测其空间结构是计算化学和计算生物物理学的重大挑战。随着计算机性能的提升，蛋白质分子力场以及增强构象取样方法的发展，近来人们已经能通过分子动力学（MD）模拟将一些快速折叠的小蛋白质折叠到接近实验的结构。然而，这种从头折叠方法还远没有成为蛋白质结构预测的实用方法。本项目将在申请人前一个项目中发展的基于显式水模型的残基特异性力场RSFF1和RSFF2的基础上，进一步发展隐式溶剂模型力场和增强构象抽样方法，实现一些折叠较慢或较大的蛋白质的准确的从头结构预测。为了大幅度地提高折叠速率，将尝试在力场中加入一些额外的相互作用来降低折叠过程的自由能垒。本项目的研究不仅能推动更好的蛋白质结构预测方法的发展，还能对蛋白质折叠机理这个重要的科学问题提供新的认识。

多肽和蛋白质的折叠与结构预测是很重要的基础科学问题。本项目在之前负责人开发的残基特异性力场（RSFF系列）的基础上开发了新的RSFF2+和RSFF2C力场。其中，RSFF2+能更好地描述蛋白折叠热力学，特别是折叠自由能的温度依赖性；RSFF2C能在各种不同的MD软件上实现，可充分利用GPU的强大计算性能。同时，本项目还利用RSFF系列力场实现了环状多肽结构的准确预测、天然无序蛋白功能的研究、小蛋白的折叠机理研究、并研究了富含二硫键小蛋白的结构预测与蛋白环区结构预测等重要课题。期间共发表学术论文7篇（均为通讯或一作），包括在物理化学领域顶级期刊JPCL上发表的两篇论文。