作者: Guixia Liu , Zengrui Wu , Yun Tang , Weihua Li , Yayuan Peng
关键词: Identification (information) 、 External validation 、 Drug repositioning 、 Drug Pathway 、 Tumor heterogeneity 、 Cellular pathways 、 Computational biology 、 Cancer therapy 、 Area under curve 、 Computer science
摘要: Identification of drug-pathway associations plays an important role in pathway-based drug repurposing. However, it is time-consuming and costly to uncover new experimentally. The drug-induced transcriptomics data provide a global view cellular pathways tell how these change under different treatments. These enable computational approaches for large-scale prediction associations. Here we introduced DPNetinfer, novel method predict potential based on substructure-drug-pathway networks via network-based approaches. results demonstrated that DPNetinfer performed well pan-cancer network with AUC (area curve) = 0.9358. Meanwhile, was shown have good capability generalization two external validation sets (AUC 0.8519 0.7494, respectively). As case study, used repurposing cancer therapy. Unexpected anticancer activities some nononcology drugs were then identified the PI3K-Akt pathway. Considering tumor heterogeneity, seven primary site-based models constructed by networks. In word, provides powerful tool A web freely available at http://lmmd.ecust.edu.cn/netinfer/.