Compound-protein interaction prediction based on heterogeneous network reveals potential antihepatoma agents

Abstract

Summary: Despite the development of an increasing number of multi-kinase and immune checkpoint inhibitors for hepatocellular carcinoma (HCC), improvement in cancer survival remains limited due to their similar structures and targets. Natural products (NPs) maintain diverse structures and activities and are important sources of drug discovery. Currently, most of active NPs exhibit ambiguous binding targets and mechanisms. Herein, we proposed the CIPHEN (compound-protein interactions prediction based on the heterogeneous network) to predict potential antihepatoma NPs and their targets. The evaluation of canonical compound-protein interactions (CPIs) databases and independent test demonstrated the good ability of CPIHN to reveal known and unreported CPIs. Both prediction and experiment results indicated that CIPHEN could unveil relationships between actively antihepatoma sesquiterpenoid dimers (SDs) and anti-HCC targets. In conclusion, the CIPHEN provides a promising choice to investigate the mode of action of compounds, which will help to accelerate the process of drug research and development against HCC.

Keywords

• Oncology
• Machine learning