Deep Docking: A Deep Learning Platform for Augmentation of Structure Based Drug Discovery

Area of Science:

• Computational chemistry
• cheminformatics
• Artificial intelligence in drug discovery

Background:

• Drug discovery is a lengthy and expensive process, often taking decades and billions of dollars.
• Virtual screening methods like molecular docking accelerate drug discovery but struggle with massive chemical databases exceeding billions of records.
• The rapid expansion of chemical libraries presents opportunities but necessitates faster screening protocols.

Purpose of the Study:

• To introduce Deep Docking (DD), a novel deep learning platform for rapid and accurate docking of billions of molecular structures.
• To address the challenge of screening vast chemical libraries in drug discovery.
• To provide a computationally efficient method for identifying potential drug candidates.

Main Methods:

• Developed Deep Docking (DD), a deep learning platform utilizing quantitative structure-activity relationship (QSAR) deep models.
• Trained QSAR models on docking scores from subsets of chemical libraries to predict outcomes for unprocessed entries.
• Iteratively removed unfavorable molecules to approximate docking outcomes.
• Integrated DD with the FRED docking program for screening.

Main Results:

• Successfully screened 1.36 billion molecules from the ZINC15 library against 12 target proteins.
• Achieved up to 100-fold data reduction and 6000-fold enrichment of high-scoring molecules.
• Demonstrated rapid and accurate calculation of docking scores without significant loss of favorable entities.
• Validated the DD methodology's effectiveness in large-scale virtual screening.

Conclusions:

• Deep Docking (DD) offers a significant advancement in accelerating drug discovery by enabling rapid screening of massive chemical libraries.
• The DD platform provides an efficient and accurate method for identifying promising drug candidates from billions of molecules.
• The DD protocol is versatile, compatible with any docking program, and has been made publicly available for broader research use.