Human Protein Complex Signatures for Drug Repositioning

Drug repositioning approaches are attracting more and more attentions in drug discovery field. Benefiting from the high-throughput gene expression data, many computational drug repositioning approaches use gene signatures to represent diseases and drugs, to identify potential drugs for diseases. Then the gene signature is used to identify potential drugs for a disease. However, the gene signatures do not take the dependencies between genes into account in the development of diseases. In this paper, we proposed human protein complex (HPC) signatures to identify potential drugs for diseases. The human protein complex (HPC) features are identified from the comprehensive resource of mammalian protein complexes (CORUM) database Based on the gene expression values, the HPC expression values are calculated. All the gene expression profiles of diseases and drug perturbations are converted to the profiles of HPCs. The HPC signatures are identified from the profiles and a list of drug candidates is generated. The results of 5 cancers indicate that the proposed method identifies more known drugs, compared with gene signature methods.