Bib | graph representation based on heterogeneous information network learning to predict drug disease association

Article address ： https://academic.oup.com/bib/article/23/1/bbab515/6456295?searchresult=1

Code address ： https://github.com/stevejobws/HINGRL

Today, I introduce Mr. Hu Lun from Xinjiang Institute of physical and chemical technology, Chinese Academy of Sciences 2022 Years published in Briefings in Bioinformatics An article on “HINGRL: predicting drug–disease associations with graph representation learning on heterogeneous information networks”. This paper presents a graph representation learning method based on Heterogeneous Information Network , Used to predict drugs - Disease related .

 

Identifying new indications of drugs plays an important role in many stages of drug research and development . Most of the existing methods complete the task by building various heterogeneous networks , Without considering the biological knowledge of drugs and diseases . So , The author proposes a heterogeneous information network based on graph representation learning (HIN) Model , namely HINGRL, For accurate identification of new indications of drugs . say concretely ,HINGRL First, by integrating drugs - disease 、 medicine - Protein and protein - The network of diseases , Combine the biological knowledge of drugs and diseases , Construct a HIN. then , Use different presentation strategies to learn from the perspective of network topology and Biology HIN Characteristics of nodes in . Last ,HINGRL use Random Forest classifier , Predict unknown drug disease associations . Experimental results show that , Compared with the most advanced model at present ,HINGRL The best performance has been achieved on two real data sets .