时 间：2019年5月29日(周三)上午10:00-11:00
地 点：中北校区理科大楼A1514会议室
题 目：Deep Learning for Loop Modeling, Contact Prediction, and Model Evaluation in Protein Structure Prediction
报告人：Shang Yi University of Missouri
摘 要：
Protein structure prediction is one of biggest unsolved problems in science and has attracted new attentions due to the success of Google’s AlphaFold in last year’s CAPS13 competition. In this talk, I will present some new deep learning methods for loop modeling, contact prediction, and model evaluation in protein structure prediction. For loop modeling, a new method based on Generative Adversarial Network (GAN) is proposed to predict the missing regions of a partially known protein structure. Experimental results based on benchmark datasets show that the new method outperformed previous methods significantly, up to 43.9%. For residue contact prediction, a new two-stage method based on convolutional neural networks and dilated residual networks is proposed. In experiments on CASP13 targets, the new method performed better than one-stage networks and existing tools. Finally, for protein model evaluation, a new method that can effectively use multiple known templates of protein fragments to improve the accuracy of evaluation result is proposed. In CASP13, this method achieved outstanding performance, ranked No. 1 in some quality assessment (QA) categories. These new methods have been implemented as web services and are freely available to the research community.
报告人简介：
Yi Shang is Professor and Director of Graduate Studies, Department of Electrical Engineering and Computer Science, University of Missouri, Columbia, Missouri. He has published over 190 refereed papers in the areas of artificial intelligence, wireless sensor networks, mobile computing, and bioinformatics and has been granted 6 US patents. He has advised over 70 PhD and MS students. His research has been supported by NSF, NIH, US Department of Education, Army, DARPA, Microsoft, Raytheon, Missouri Department of Conservation, etc.