Local sequence alignment algorithms and software tools using the next generation hybrid CPU/GPU parallelism.

摘要

Sequence Alignment is an important problem in computational sciences with a wide variety of applications. For example, in the field of historical and comparative linguistics, sequence alignment has been used to partially automate the comparative method by which linguists traditionally reconstruct languages. In business and marketing research, multiple sequence alignment techniques has been applied in analyzing series of purchases over time. Especially, sequence alignments are very useful in bioinformatics for identifying DNA, RNA and protein sequence similarity, producing phylogenetic trees, and developing homology models of protein structures. However, with the fast development of the technology for high throughput sequencing, databases being searched have been increased exponentially in recent years. Therefore, the need for new algorithms, which can allow researchers to search for sequence matching in an efficient manner has become urgent. The main idea of this thesis is to improve the performance of existed Local Sequence Alignment Search Algorithms and Tools by utilizing the High Performance computing capabilities with the help of CUDA programming language.