“…Since the binding cavity of COVID-19 3CLpro and SARS 3CLpro is extremely similar ( Supplementary Figure S1 ), it is rational to make the learning model based on the data of inhibitors of above two proteins. The 66 active compounds and 66 inactive compounds were collected as the training and testing set data (patent: US7495011 B2) ( Yamamoto et al, 2004 ; Chen et al, 2006 ; Wang et al, 2007 ; Jacobs et al, 2013 ; Adedeji and Sarafianos, 2014 ; García-Fernández et al, 2016 ; Konno et al, 2017 ; Hu et al, 2020 ; Jo et al, 2020 ; Yoshizawa et al, 2020 ). The chemical structures of compounds were drawn by ChemDraw software and translated into canonical SMILES by the RDKit python package ( Shi and Borchardt, 2017 ).…”