Kanserde ilaç etkilerinin ve benzerliklerinin bulunması amaçlı çoklu omik veri entegrasyonu ile biyolojik ağ modelleme

Classification of cancer drugs is crucial for drug repurposing since the cost and innovation deficit make new drug development processes challenging. Heterogeneity of cancer causes drug classification purely based on known mechanism of action (MoA) and the list of target proteins to be insufficient. Multi-omic data integration is necessary for a systems biology perspective to understand molecular mechanisms and interactions between cellular entities underlying the disease. This study integrates drug-target interaction data with transcriptomics and phosphoproteomic data of perturbed cell lines to model drug and cell-specific subnetworks. Total 250 networks are reconstructed, including 70 small molecule drugs on six cell lines. Similarities of reconstructed networks are quantitatively calculated using a topology-based network comparison measure which scores the separation of networks using the shortest paths between network nodes. Different drugs with similar omic outcomes on variable cell lines are revealed with the aid of separation scores. Moreover, the effect of drugs on variable cell lines is discovered together with the impact of target selectivity of drugs within the same MoA group. Functional analysis of reconstructed networks for their enriched cellular pathways further indicated that drugs with different chemical structures and MoA might induce common signaling cascades. As omics data integration coupled network modeling reveals modulated pathways for specific conditions, the methodology of this study is applicable to different drug-disease research areas. Prediction of drug combinations for a given disease and inference of drug similarity based on cell line sensitivity are two applications presented in this study.