1. FRODRUG: A Virtual Screening GPU Accelerated Approach for Drug Discovery
- Author
-
Pablo Chacón, Erney Ramírez-Aportela, Raúl Cabido, José Ignacio Garzón, Santiago Garcia, and Antonio S. Montemayor
- Subjects
Instruction set ,Multi-core processor ,Virtual screening ,Drug discovery ,Computer science ,Docking (molecular) ,Parallel algorithm ,Single-core ,Algorithm design ,Parallel computing - Abstract
The procedure for screening large databases of small chemical compounds to select likely drug candidates by computational means is very time demanding. Here, we present and evaluate a new method for virtual screening (VS) that combines the efficiency of spherical harmonic approximations to accelerate the rotational part of a docking search with multicore and GPU parallelism. To validate these novel parallel algorithms, we used standard benchmark cases. The obtained results are comparable to those generated via state-of-the-art VS docking approximations, but with a considerable gain in efficiency. GPU implementation speedups of more than 30-fold with respect to a single core CPU were achieved, reducing the docking time for a single ligand to only 50 milliseconds. The achieved efficiency and the accuracy on standard blind benchmarks demonstrate the applicability and robustness of this approach.
- Published
- 2014
- Full Text
- View/download PDF