UniDock-Pro: A Unified GPU-Accelerated Platform for High-Throughput Structure-Based, Ligand-Based, and Synergistic...

Area of Science:

• Computational chemistry
• Drug discovery
• Bioinformatics

Background:

• High-throughput computational tools are essential for screening ultralarge chemical libraries.
• Existing methods require efficient exploitation of structural and chemical information.

Purpose of the Study:

• To present UniDock-Pro, a unified platform for accelerated virtual screening.
• To integrate structure-based virtual screening (SBVS), ligand-based virtual screening (LBVS), and a novel hybrid mode.
• To enhance computational efficiency and accuracy in drug discovery.

Main Methods:

• Developed UniDock-Pro, a GPU-accelerated platform based on Uni-Dock architecture.
• Implemented enhanced LBVS with a smooth potential for gradient-based search.
• Introduced a hybrid screening mode integrating receptor- and ligand-derived grid maps.
• Developed Force Field Complementarity Analysis (FFCA) to quantify spatial alignment.

Main Results:

• UniDock-Pro processes millions of compounds per day on a single GPU.
• Achieved a 2.42-fold improvement in early enrichment (EF1%) over AutoDock-SS on the DUDE-Z benchmark.
• Demonstrated strong early enrichment on DUDE-Z and competitive performance on VSDS-vd TrueDecoy benchmark using Hybrid mode.

Conclusions:

• UniDock-Pro offers a robust, versatile, and highly efficient solution for drug discovery.
• The platform significantly accelerates campaigns across large chemical spaces.
• Enhanced LBVS and the novel hybrid mode improve virtual screening accuracy and speed.