学术讲座
报告题目：Computational protein design and stability prediction using neural networks
报告人：戚逸飞 博士
华东师范大学
时间：2019年04月22日（周一）14:30
地点：曾呈奎楼3楼B311
嘉宾介绍：
戚逸飞博士2006和2012年分别在北京大学获得学士和博士学位。12-16年在美国堪萨斯大学进行博士后研究。16年之后加入华东师范大学化学与分子工程学院，目前为副研究员。他的主要研究方向是生物大分子的结构和功能模拟以及蛋白质设计等。目前已经发表论文近40篇，总引用1000余次。
报告摘要：
Computational protein design has a wide variety of applications. Despite its remarkable success, designing a protein for a given structure and function is still a challenging task. Deep learning neural network is a powerful method to learn big data set and has shown superior performance in many machine learning tasks. We have applied the deep learning neural network approach to computational protein design for predicting the probability of 20 natural amino acids on each residue in a protein. A large set of protein structures was collected and multi-layer neural network was constructed. A number of structural properties were extract as input features and the best network achieved an accuracy of 38.3%. Using the network output as residue type restraints was able to improve the average sequence identity in designing three natural proteins using Rosetta. We have also adapted the protein design neural network to predict the stability change of protein upon point mutations. The neural network was trained on more than 5700 manually curated experimental data points and was able to obtain a Pearson correlation coefficient of 0.48-0.56 for three independent test sets, which outperformed eleven other methods. Detailed analysis of the input features shows that the solvent accessible surface area of the mutated residue is the most important feature, suggesting that the buried hydrophobic area is the major determinant of protein stability. Both neural networks are freely available at http://protein.org.cn.
