Real-Time Protein-Protein Docking Using GPU and Machine Learning

EasyChair Preprint 14199

Abstract

Real-time protein-protein docking is a critical task in computational biology, crucial for understanding molecular interactions and designing targeted therapeutics. Traditional docking methods, while effective, often struggle with the computational complexity and time constraints inherent in processing large-scale protein interactions. This study explores the integration of Graphics Processing Units (GPUs) and machine learning (ML) techniques to enhance the efficiency and accuracy of protein-protein docking in real time. By leveraging the parallel processing capabilities of GPUs, we accelerate the docking simulation process, allowing for the rapid assessment of potential protein interactions. Additionally, machine learning models are employed to predict binding affinities and optimize docking configurations, improving the predictive power and reducing computational overhead. Our approach involves the development of a GPU-accelerated docking algorithm, combined with a ML-driven scoring function that adapts to various protein complexes. Preliminary results demonstrate significant improvements in processing speed and accuracy compared to traditional methods. This advancement promises to facilitate more dynamic and detailed studies of protein interactions, with potential applications in drug discovery and biomolecular research.

Keyphrases: Graphics Processing Units (GPUs), Machine Learning (ML), Protein-protein interactions (PPIs

@booklet{EasyChair:14199,
  author    = {Abill Robert},
  title     = {Real-Time Protein-Protein Docking Using GPU and Machine Learning},
  howpublished = {EasyChair Preprint 14199},
  year      = {EasyChair, 2024}}