GSA: a GPU-accelerated structure similarity algorithm and its application in progressive virtual screening.

Area of Science:

• Computational chemistry
• Bioinformatics
• Drug discovery

Background:

• Chemical structure similarity search is crucial for virtual screening in drug discovery.
• Current methods utilizing central processing units (CPUs) can be computationally intensive and slow.
• Accelerating these searches is essential for efficient identification of potential drug candidates.

Purpose of the Study:

• To develop and evaluate a new algorithm for accelerating chemical structure similarity searches.
• To leverage graphic processing unit (GPU) technology for enhanced computational speed.
• To assess the algorithm's effectiveness in progressive virtual screening experiments.

Main Methods:

• Development of a novel algorithm for chemical structure similarity search.
• Implementation of the algorithm on GPU hardware.
• Experimental comparison against conventional CPU-based algorithms.
• Application of the algorithm in seven virtual screening experiments using NCI/DTP 60 human cancer cell lines data.

Main Results:

• The proposed GPU-based algorithm demonstrated an average speedup of over 120 times compared to CPU-based methods.
• The algorithm successfully identified 70-80% of intrinsic hits by screening only 10-20% of compounds in virtual screening experiments.
• The approach proved generalizable across multiple virtual screening scenarios.

Conclusions:

• GPU technology offers a significant advantage for accelerating chemical structure similarity searches.
• The developed algorithm substantially improves the efficiency of virtual screening processes.
• This technology has the potential to expedite the identification of novel therapeutic compounds.