| 日時(Date) | 平成30年10月25日（木）3限（13:30--15:00) Thur. Oct. 25th, 2018, 3rd Period (13:30--15:00) |
|---|---|
| 場所(Location) | L1 |
| 司会(Chair) | Assoc. Prof. Md.Altaf-Ul-Amin |
| 講演者(Presenter) | Dr. Pingzhao Hu ( Department of Biochemistry and Medical Genetics, University of Manitoba, Canada) |
| 題目(Title) | Predicting drug - target interaction network using deep learning models for drug repurposing through genetic information |
| 概要(Abstract) | Problem: Traditional methods for drug discovery are time-consuming and expensive, so new efforts have been made to perform drug repurposing, which can develop new uses from already approved drugs. In order to explore new ways for this innovation, many computational approaches have been proposed to predict drug-target interactions (DTIs). However, due to the high-dimensional nature of the data sets extracted from drugs and targets, traditional machine learning approaches such as logistic regression analysis could not analyze these data efficiently. In this thesis, we aim to propose LASSO-based regularized linear classification models and a deep neural network (DNN) model to predict DTIs and apply clustering analysis to the new predictions for exploring potential drugs in enriched clusters for inflammatory bowel disease (IBD) and breast cancer. Experimental results showed that the DNN model outperformed the SLR and LASSO models. Particularly, LASSO models with protein tripeptide composition (TC) features and domain features were superior to those that contained other protein information in LASSO models, which may imply that TC and domain information could be better representations of proteins. Furthermore, we showed that using DTIs predicted by the DNN model, the potential drugs we collected for IBD and breast cancer from the enriched clusters can potentially be repurposed for IBD and breast cancer treatment. |
| 講演言語(Language) | English |
| 講演者紹介(Introduction of Lecturer) | Dr. Pingzhao Hu is a tenure-track Assistant Professor in Bioinformatics in The Department of Biochemistry and Medical Genetics and The Department of Computer Science at The University of Manitoba and Assistant Professor (Status) in Division of Biostatistics, Dalla Lana School of Public Health, University of Toronto. Dr. Hu received his PhD in computer science from York University. He pioneered machine learning algorithms for protein function predictions, which were published on Plos Biology and Nature Reviews Cancer. His group develops and applies novel deep learning and statistical algorithms for integrative analysis of high-throughput multimodal (genomic and imaging, etc) data to understand the causal associations between human genome and phenome. Dr. Hu has published over 80 peer reviewed scientific articles with over 3100 citations. His research program is funded by NSERC Discovery, NSERC CREATE, Canada Breast Cancer Foundation (CBCF), Mitacs, Canadian Institute for Health Research (CIHR), Research Manitoba, etc. He is a recent receipt of The Interstellar Initiative Award funded by New York Academy of Sciences and Japan Agency for Medical Research and Development. |