Research On Parallel Processing Technology Of Sequence Analysis

Bioinformatics is a new science field. Research in this field involves multi-disciplines such as biology, computer science, mathematics, etc. Bioinformatics is subject to expose the biological signification of large amount of biological data and explore the mystery of life activities. To improve the processing performance of biological sequence is an important task in the research of bioinformatics. The key point of this research is to develop more efficient parallel algorithm of sequence processing with different computing environment. Much more work on the fields had been taken and some fruits had been acquired.After analyzing existing sequence processing methods and software, in order to improve the parallel scalability and speed,some novel parallel algorithms for sequence processing on different computing environment, such as shared memory architecture,distributed memory architecture and grid computing architecture, are presented in this thesis.Firstly, to improve the mpiformatdb's speed, a novel parallel algorithm based on shared memory architecture is presented. By adding OpenMP directives, the cycled parallel structure is formed from the serial algorithm. Secondly, to improve the hmmsearch's parallel scalability, a new two-level parallel algorithm based on sequence partition strategy is presented, which is a dynamic load balancing scheme. By introducing the intermediate layer, a new"Master-subMaster-Slave"structure is better than the"Master-Slave"structure in parallel scalability. Finally, in order to improve the BAC's assembly speed, a novel BARNACLE parallel algorithm is presented. And based on the Globus Toolkit software packet, the parallel algorithm is implemented in the grid computing environment.Some experiments of all these parallel algorithms are presented,the testing results indicate that the proposed algorithms are of high efficiency.