High throughput TCR sequence alignment using multi-GPU with inter-task parallelization

Based on GPU computing, a fast computing using multi-GPU is proposed for the alignment of vast amounts of T-cell receptor (TCR) nucleotide sequences. Using CUDA-enabled Fermi GPU and CUDA toolkit 4.0 provided by NVIDIA, we design a faster and more effective sequence alignment process based on CPU an...

Full description

Saved in:
Bibliographic Details
Main Authors: Guoli Ji, Qiang Li, Mingcheng Wu, Jingyi Fu, Xiaorong Hu, Liangwang Chi, Qi Liu
Format: Conference Proceeding
Language:English
Subjects:
CUDA
Multi-GPU
Sequence alignment
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Based on GPU computing, a fast computing using multi-GPU is proposed for the alignment of vast amounts of T-cell receptor (TCR) nucleotide sequences. Using CUDA-enabled Fermi GPU and CUDA toolkit 4.0 provided by NVIDIA, we design a faster and more effective sequence alignment process based on CPU and multi-GPU: CPU is responsible for logic control, GPU responsible for parallel computing. Inter-task parallel strategy is applied in the part of parallel computing, which not only bring high parallelism, but also make the alignment process not confined to a specific parallel alignment algorithm. Under the same hardware condition, the alignment computing of mouse TCR nucleotide sequences were carried out by multi-GPU computing, single-GPU computing and only-CPU computing respectively. The results show that multi-GPU computing has the best performance considering alignment efficiency and the cost.
DOI:10.1109/IECBES.2012.6498184