Machine Learning Models Identify Inhibitors of SARS-CoV-2

• DOI: 10.1021/acs.jcim.1c00683
• Publication Date: 2021-08-13T19:33:52Z
• Authors: Victor O. Gawriljuk; Phyo Phyo Kyaw Zin; Ana C. Puhl; Kimberley M. Zorn; Daniel H. Foil; Thomas R. Lane; Brett Hurst; Tatyana Almeida T...; Fabio Trindade Ma...; Premkumar Lakshma...; Jean Bernatchez; Andre S. Godoy; Glaucius Oliva; Jair L. Siqueira-...; Peter B. Madrid; Sean Ekins
• Keywords: Microscale Thermophoresis; Repurposing; Drug repositioning; In vitro toxicology; Coronavirus; Computational model

With the rapidly evolving SARS-CoV-2 variants of concern, there is an urgent need for discovery further treatments coronavirus disease (COVID-19). Drug repurposing one most rapid strategies addressing this need, and numerous compounds have already been selected in vitro testing by several groups. These led to a growing database molecules with activity against virus. Machine learning models can assist drug through prediction best based on previously published data. Herein, we implemented machine methods develop predictive from recent inhibition data used them prioritize additional FDA-approved our in-house compound library. From predicted Bayesian model, lumefantrine, antimalarial was showed limited antiviral cell-based assays while demonstrating binding (Kd 259 nM) spike protein using microscale thermophoresis. Several other which prioritized since tested others were also found be active vitro. This combined approach expanded virtually screen available reference WIV04 strain circulating concern. In process work, created multiple iterations that as prioritization tool programs. The very latest model over 500 now freely at www.assaycentral.org.