Extracting information on virus-human interactions and on antiviral compounds based on automated analysis of large text collections.

The development of effective antivirals is of great importance due to the threat associated with the rapid spread of viral infections. The accumulation of data in scientific publications and in databases of biologically active compounds provides an opportunity to extract specific information about interactions between chemicals and their viral and host targets. This information can be used for elucidation of knowledge about potential antiviral activity of chemical compounds, their side effects and toxicities. Our study aims to extract knowledge about virus-host interactions and potential antiviral agents based on the mining of massive amounts of scientific publications. With a set of previously developed algorithms, we have extracted comprehensive information on virus-host interactions and chemical compounds that interact with both viral and host targets. We collected data on the interactions of several viruses, including hepatitis B and C viruses, SARS-CoV-2, influenza A and B, and herpes simplex viruses, with (1) the host (human body), (2) potential antiviral agents, and, also extracted information on the interactions between potential antiviral agents and host proteins. Based on the data analysis performed, we created a freely available knowledge base on the interaction of chemical compounds with viral proteins and their host targets, allowing the exploration of both well-studied and recently discovered novel virus-host-chemical-compound interactions.