A Large-scale Drug Repositioning Survey for SARS-CoV-2 Antivirals

Laura Riva; Shuofeng Yuan; Xin Yin; Laura Martin-Sancho; Naoko Matsunaga; Sebastian Burgstaller-Muehlbacher; Lars Pache; Paul P. De Jesus; Mitchell V. Hull; Max Chang; Jasper Fuk-Woo Chan; Jianli Cao; Vincent Kwok-Man Poon; Kristina Herbert; Tu-Trinh Nguyen; Yuan Pu; Courtney Nguyen; Andrey Rubanov; Luis Martinez-Sobrido; Wen-Chun Liu; Lisa Miorin; Kris M. White; Jeffrey R. Johnson; Christopher Benner; Ren Sun; Peter G. Schultz; Andrew Su; Adolfo Garcia-Sastre; Arnab K. Chatterjee; Kwok-Yung Yuen; Sumit K. Chanda

The dynamic range of the viral-induced CPE in the assay was small (~2-2.5 fold), but robust and reproducible ( Figure 1B) . Both the optimization using the LOPAC® 1280library and the first ReFRAME collection screen displayed acceptable Zʹ factors (0.4 and 0.51, respectively). The duplicate ReFRAME screen, had a reduced dynamic range (1.5-fold) and corresponding Zʹ factor (0. 19) . Although the correlation between the two ReFRAME replications was high (R 2 =0.68), there were compounds that were found active in replicate 1, but not replicate 2 ( Figure 1C , bottom right of middle panel). While we leveraged all datasets to select molecules for further validation, compound selection was weighted for replicate 1. Specifically, in addition to 28 molecules from the LOPAC®1280 library, 250 drugs were selected based on their activity. 48 additional ones, belonging to enriched target/MOA sets based on GSEA analysis, were also included.