Cancer-Drug Interaction Network Construction and Drug Target Prediction Based on Multi-source Data

With the finish of the human genome sequencing and the great progress in molecular biology like proteomics, many established authoritative international biomedical databases are completing continually in recent years. With these opening databases, all kinds of biological molecular networks can be constructed for potential disease gene detection and drug target prediction through network-based approaches. However, most methods do the drug target prediction along with data from only a single source, which have many limitations and tendencies. In this paper, we use multi-source data integrate with datasets from Uniprot, HGNC, COSMIC and DrugBank to do the anti-cancer drug target prediction more comprehensively. We construct Drug-Target network (DT network), Cancer-Gene network (CG network) and Cancer-Drug Interaction network (CDI network) based on the multi-source data we integrate, and do visualizations of the three networks in Cytoscape. In addition, we make an anti-cancer drug target prediction with the method of Random Walks on graphs, one of the most efficient method in biological molecular network analysis by now. Potential anti-cancer drug targets are predicted by calculating the correlation strengths between known cancer gene products and other proteins in CDI network with PersonalRank algorithm. Analysis of the prediction results shows that the potential anti-cancer drug targets we predicted are highly related with cancers both topologically and bio-functionally, which verifies the rationality and availability our method.