Аннотация:Emergence of novel viruses, such as Ebola virus and Zika virus, and resistance of known ones, such as influenza virus, HCV, and HIV, justifies discovery and design of new antiviral drugs as a very important branch of medicinal chemistry. Drug discovery based on previously obtained data is a widely accepted approch. The most widely used public repository ChEMBL provides access to a large amount of antiviral activity data, but these data are often insufficiently annotated and poorly curated. To overcome this problem, we developed an algorithm of semi-automatic curation of ChEMBL data based on mapping lists for assay organism and target organism data and a dictionary of virus-related terms. With the help of this algorithm ChEMBL 20 and ICTV taxonomy 2014 were used for the generation of the first version of antiviral activity database ViralChEMBL, which provided the most comprehensive and the best annotated antiviral activity profiles for small molecule compounds to date. Applicability of ViralChEMBL for antiviral chemical space mapping was illustrated using the approach of self-organised maps.