Fast Index Construction Of Biological Data Using Graphics Processing Unit (GPU)
Loading...
Date
2014-07
Authors
Rahmaddiansyah
Journal Title
Journal ISSN
Volume Title
Publisher
Universiti Sains Malaysia
Abstract
Technological advancement is producing terabytes of biological data in the
scientific community. Such data include deoxyribonucleic acid and protein
sequences. String matching algorithm was traditionally used to match DNA
sequences but requires a long time to execute because of the large size DNA and
protein sequences. Indexing methods, such as suffix arrays (SA) and suffix trees
(ST), have been introduced to remedy the processing problem. In this study, SA are
used as an indexing algorithm because such SA are more applicable than ST. SA are
not complex and are space efficient. The parallel method is introduced to accelerate
the index construction process. Graphics Processing Unit is a processor used to
parallelize a part of an indexing algorithm. This work shows how the GPU is used to
sort parts of the SA construction algorithm. Results show that the GPU can
accelerate the construction of SA. The use of a GPU can process DNA data 2.01
times faster and protein data 1.63 times faster than the case without a GPU.
Description
Keywords
Indexing technique