Vina-GPU 2.0

Vina-GPU 2.0 accelerates molecular docking to enable high-throughput virtual screening and candidate prioritization in drug discovery.

Key Features:

• GPU-optimized docking algorithms: Implements GPU-optimized variants of AutoDock Vina derivatives, including QuickVina 2 and QuickVina-W.
• Acceleration strategies: Adapts different acceleration strategies tailored to each algorithm to exploit GPU parallelism on hardware such as the NVIDIA RTX 3090.
• Performance benchmarks: Demonstrates average speedups of 65.6-fold over AutoDock Vina, 1.4-fold over QuickVina 2, and 3.6-fold over QuickVina-W while maintaining comparable docking accuracy.
• Benchmark dataset and targets: Evaluated by virtual screening of protein kinases RIPK1 and RIPK3 using compounds from the DrugBank database.

Scientific Applications:

• Large-scale virtual screening: Reduces computational time to enable screening of large compound libraries such as DrugBank while preserving docking accuracy.
• Kinase-targeted screening: Applied to prioritize potential therapeutic compounds against protein kinases RIPK1 and RIPK3.

Methodology:

Optimizes existing docking algorithms (AutoDock Vina derivatives including QuickVina 2 and QuickVina-W) for GPU execution by adapting algorithm-specific acceleration strategies to fully exploit the parallel processing capabilities of modern GPUs such as the NVIDIA RTX 3090.