Designing Potent HIV‑1 Protease Inhibitors Using Machine Learning and QSAR Approaches

Area of Science:

• Medicinal Chemistry
• Computational Biology
• Drug Discovery

Background:

• Acquired Immune Deficiency Syndrome (AIDS), caused by Human Immunodeficiency Virus type-1 (HIV-1), presents a persistent global health challenge.
• Drug resistance, especially to protease inhibitors like Darunavir, limits the effectiveness of current antiretroviral therapies against HIV-1 variants.

Purpose of the Study:

• To design novel, potent HIV-1 protease inhibitors by integrating machine learning (ML) and quantitative structure-activity relationship (QSAR) modeling.
• To identify key molecular descriptors and structural features contributing to inhibitory activity against HIV-1 protease.
• To evaluate the binding potential of newly designed inhibitors against wild-type and resistant HIV-1 strains.

Main Methods:

• Development and comparison of multiple QSAR models (GFA, MLR, RF, GBR, XGBoost) using various molecular descriptors.
• Application of SHAP analysis to interpret model predictions and identify critical molecular features influencing pIC50.
• Structure-based molecular docking simulations to assess binding affinity and interactions of predicted inhibitors with HIV-1 protease.

Main Results:

• Gradient Boosting Regressor (GBR) models achieved high accuracy (R² = 0.911 and 0.994) in predicting inhibitory activity.
• Key predictive features included electronic charge at C53, low dipole moments, and a short C53-O54 bond length.
• Five potent inhibitors (B01-B05) were designed, with B03 and B05 showing strong predicted binding to wild-type and variant HIV-1 proteases via hydrophobic and hydrogen bonds.

Conclusions:

• The integrated QSAR-ML and structure-based approach successfully identified promising novel HIV-1 protease inhibitor candidates.
• The designed compounds demonstrate potential for overcoming existing drug resistance mechanisms in HIV-1 treatment.
• This study provides a foundation for developing next-generation antiretroviral therapies against resistant HIV-1 strains.