DrugPipe: Generative artificial intelligence-assisted virtual screening pipeline for generalizable and efficient drug...

Area of Science:

• Computational chemistry
• Drug discovery and development
• Artificial intelligence in medicine

Background:

• Drug repurposing offers a faster route to new therapeutics, especially for diseases lacking effective treatments.
• Identifying novel therapeutic applications for existing drugs is crucial for unmet medical needs.
• Target-centric drug repurposing focuses on specific protein targets to find interacting compounds.

Purpose of the Study:

• To introduce DrugPipe, a Generative AI-Assisted Virtual Screening Pipeline for target-centric drug repurposing.
• To enable scalable and generalizable in silico drug repurposing workflows.
• To support blind virtual screening for any protein target, including novel or understudied ones.

Main Methods:

• DrugPipe integrates generative modeling, binding pocket prediction, and similarity-based retrieval from drug databases.
• The pipeline supports blind virtual screening without prior structural or functional annotations for protein targets.
• It efficiently generates candidate ligands and retrieves structurally similar approved drugs.

Main Results:

• DrugPipe achieves hit rate performance comparable to established tools like QVina-W.
• The pipeline significantly reduces computational time compared to traditional virtual screening methods.
• It demonstrates practical value for large-scale virtual screening and data-scarce repurposing scenarios.

Conclusions:

• DrugPipe accelerates the identification and prioritization of repurposable compounds for various diseases.
• The AI-assisted pipeline is particularly valuable for novel targets and emerging health threats.
• The developed workflow enhances the efficiency and scalability of drug repurposing efforts.