An Interpretable Deep Learning and Molecular Docking Framework for Repurposing Existing Drugs as Inhibitors of...

Area of Science:

• Computational Chemistry
• Drug Discovery
• Bioinformatics

Background:

• COVID-19 remains a global health challenge despite widespread vaccination.
• Efficient drug screening and repurposing are crucial for developing new therapeutics.
• Existing methods for identifying potential drug candidates can be time-consuming.

Purpose of the Study:

• To develop and validate a novel hybrid framework integrating deep learning and molecular docking.
• To accelerate the identification of potential therapeutic agents for COVID-19.
• To identify inhibitors of the SARS-CoV-2 main protease (MPro).

Main Methods:

• A deep learning model was used for initial rapid screening of candidate compounds.
• Molecular docking tools (AutoDock Vina, LeDock) evaluated binding affinities.
• Predicted drug-protein binding sites were assessed for overlap with known active residues.

Main Results:

• The framework was validated using four experimentally confirmed COVID-19 drug-target pairs.
• Enasidenib was identified as a promising MPro inhibitor among 29 drug candidates.
• Enasidenib met all three selection criteria, including predicted interaction score, binding affinity, and binding site overlap.

Conclusions:

• The hybrid deep learning and molecular docking framework offers an efficient strategy for virtual screening and drug repurposing.
• Enasidenib shows potential as an MPro inhibitor, but requires further experimental and clinical validation.
• This interpretable strategy can be adapted for other drug discovery targets and computational tools.