A Sequence Alignment Strategy for Large-Scale Biological Sequences Using GPU

• Main Authors: Pai-Chen Su, 蘇柏丞
• Other Authors: Hsing-Lung Chen
• Format: Others
• Language: zh-TW
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/76166879648626876960

碩士 === 國立臺灣科技大學 === 電子工程系 === 104 === In bioinformatics, a sequence alignment is the most basic and quite important research tool. It is a way of identifying regions of similarity that may be a consequence of functional, structural, or evolutionary relationships between the sequences. We can speculate the sequences whether or not derived from a common ancestor. In sequence alignment, we often need many times to compare the results. Typically the length of the two input strings are very long, so the long processing time is required. There are many algorithms to reduce the processing time like Smith-Waterman. Smith-Waterman algorithm is a local alignment comparison method, but its time and space complexity is O(n²). This algorithm would not work when we alignment of two DNA sequences. Because this algorithm need a large amount of memory and disk space. It is unable to save all information and needs huge processing time. Based on Smith-Waterman and Needleman-Wunsch algorithm, we use GPU acceleration and parallel processing. The use of CUDA streams to launch kernels[1]. We repeatedly cut the entire matrix then individual operation. We let the processing time as short as possible.