8-5 GraphSynergy:用于抗癌药物组合预测的网络启发深度学习模型.pdf

《8-5 GraphSynergy:用于抗癌药物组合预测的网络启发深度学习模型.pdf》由会员分享，可在线阅读，更多相关《8-5 GraphSynergy:用于抗癌药物组合预测的网络启发深度学习模型.pdf（35页珍藏版）》请在三个皮匠报告上搜索。

1、GraphSynergy:Network inspired deep learning model for anti-cancer drug combination predictionQingpeng Zhang 张清鹏Joint work with Jiannan Yang,Zhongzhi Xu,William K.K.Wu and Qian ChuSchool of Data ScienceCity University of Hong KongAI in drug discovery and repurposingAI in drug discovery and repurposin

2、gMarinka Zitnik,Tutorial on Machine Learning for Drug Development,IJCAI21Challenges The combinatorial space of drug combinations is huge.In vitro experiments are costly.The effects of drug combination can be adverse.1+1 2?Black-box State-of-the-art deep learning methods(DeepSynergy,AuDNNsynergy):Sim

3、ilar chemical structures similar treatment effects.Ignoring the complex biological interactions among the proteins related to drugs and diseases.Network science approach The protein-protein interaction(PPI)network serves as a“skeleton”for bodys signaling circuitry.Barabsis team,Nature Communications

4、,2016&2019Network science approach An effective drug should target the proteins within or near the corresponding disease module.Two drugs with synergistic effects should target complementary(non-overlapping)proteins to prevent the toxicities caused by over-exposure.Barabsis team,Nature Communication

5、s,2016&2019Network science approach Only focus on the topological distance between proteins directly associated with drugs and diseases,while ignoring the local connections(formed by neighboring proteins)and the global structure of the PPI network.Existing network science approaches treat each prote

6、in homogeneously,whereas recent studies reveal that several proteins have a dominant contribution to the progression of cancers.Not optimized.This research We propose a novel end-to-end machine learning framework,namely Graph Convolutional Network for Drug Synergy(GraphSynergy),to identify synergist