Ai-Based QSAR Approach for Predicting Cathepsin L Inhibition

Abstract

Cathepsin L (CatL), belongs to cysteine protease class of enzymes that is primarily involved in proteolytic process in lysozymes. The enzyme levels are abnormally enhanced in several disease conditions. Recent studies have indicated Cathepsin L as a promising target to treat COVID-19. The entry of causative virus SARS-Co-V-2 into the host cell can be decreased by CatL inhibition. Several naturally occurring agents such as triperpenoids and Gallinamides inhibit CatL enzyme. Statistically significant and robust QSAR models establish correlation between molecular properties and biological activity of a set of chemical compounds. The present work deals with the development of a QSAR) model(s) for identifying relationship between the structural features and the biological activity of Cat L inhibitors. Classification-based (LDA) and regression-based (MLR) QSAR models were generated using different AI tools. [MLR-based QSAR model: Internal Validation Parameters: R² = 0.683,  R² (adjusted)= 0.663, PRESS =119.64  Cross-validation results (Leave-One-Out) : Q² :0.637 External Validation Parameters: r² :0.604,  Q²f1/R²(Pred) :0.61, Q²f2 :0.601, RMSEP:0.839]. The LDA model also passed all the qualitative validation metrics computed while performing internal and external validation using the training and test sets, respectively. The derived QSAR models are significant for predicting the activity of newly designed CatL inhibitors.