Quantitative Structure-Activity Relationship Study of Cathepsin L Inhibitors as SARS-CoV-2 Therapeutics Using...

Area of Science:

• Medicinal Chemistry
• Computational Drug Discovery
• Virology

Background:

• Cathepsin L (CatL) is a protease crucial for SARS-CoV-2 spike protein cleavage and viral entry.
• Inhibiting CatL presents a promising therapeutic strategy against SARS-CoV-2 infections.
• Developing effective CatL inhibitors requires robust predictive modeling.

Purpose of the Study:

• To develop and validate Quantitative Structure-Activity Relationship (QSAR) models for predicting CatL inhibitory activity.
• To identify novel CatL inhibitors with potential therapeutic applications against SARS-CoV-2.
• To explore the utility of various machine learning algorithms in drug discovery for viral targets.

Main Methods:

• Six QSAR models were constructed using heuristic methods (HMs), gene expression programming (GEP), random forest (RF), and support vector regression (SVR) with different kernels (RBF, LMIX2-SVR, LMIX3-SVR).
• Particle swarm optimization was employed to optimize multi-parameter SVM models.
• Molecular docking and Property Explorer Applet (PEA) were used to assess novel inhibitor properties.

Main Results:

• The LMIX3-SVR model demonstrated superior performance, achieving R² values of 0.9676 (training) and 0.9632 (test), with low RMSE.
• Cross-validation (5-fold and leave-one-out) confirmed the model's strong predictive ability and robustness.
• QSAR models predicted IC50 values for 578 novel compounds, with top candidates identified and verified.

Conclusions:

• The LMIX3-SVR model offers a reliable tool for QSAR modeling in drug discovery, significantly advancing the design of new molecules.
• This study successfully identified potential novel CatL inhibitors, contributing to the development of effective SARS-CoV-2 therapeutics.
• The findings underscore the importance of computational approaches in accelerating the identification of antiviral drug candidates.