Computational drug discovery for COVID-19 (COVID19CADD)

Abstract

The COVID-19 pandemic, caused by the novel virus SARS-CoV-2, is most likely here to stay with us and notwithstanding the rapid deployment of vaccines, there is a need to develop antivirals against this virus. This is because new variants of the virus will emerge against which current vaccines might be less effective, there will be people that cannot be vaccinated or areas of the world where deployment of vaccination programs are not as efficient as in other more developed countries. For this reason there is a significant effort to develop antivirals effective against this virus. Virtual screening is a set of computational tools within the Computer-Aided Drug Design toolbox that is available in order to perform initial filtering of small molecules in order to identify hits that show potential and merit being studied further as part of a drug development pipeline. In this study we make use of two Ligand-Based Virtual Screening (LBVS) techniques – Molecular Fingerprint Similarity Searches and Ultrafast Shape Recognition with CREDO Atom Types (USRCAT) – to search for small molecules that are similar to a set of query molecules that have been identified as having an inhibitory effect against the Main Protease (Mpro) of SARS-CoV-2. Our experiments have resulted in a list of 42 and 195 hits identified from Fingerprint and USRCAT searches, respectively. These results were validated by the calculation of the Enrichment Factor, which resulted in scores (well) above a value of 1, with mean EF1% values of 7.56 and 2.57 for Fingerprint and USRCAT searches, respectively. These values can be studied using other in silico tools such as molecular docking or in vitro studies.