Implementation of Parallel Search Algorithm in Computational Biology

Bioinformatics and Computational Biology are rapidly growing multidisciplinary fields, which includes wide variety of domains from DNA sequencing to sequence alignments. Recent advances in both these disciplines have allowed biologists all around the world to quickly gather a huge amount of DNA sequence data for analysis. DNA sequence alignments are becoming ever more popular due their impact in early disease diagnosis, in drug engineering, as well as in criminal investigations. With the vast growth and popularity of biological data, searching for a DNA sequence of interest in huge databases is not an easy task to produce results within a realistic time, hence there is a need to enhance the efficiency. The reason why such information is so popular is because biologists can identify genetic information by finding sequences of similar genes or proteins with known behavior or structure without requiring long and expensive laboratory experiments. One of the most widely used tools for performing searches is Basic Local Alignment Search Tool (BLAST), a program for performing pairwise sequence alignments. As the BLAST program becomes ever more popular with biologists around the world, it faces numerous challenges. One of the main challenges is the issue of performance. The BLAST program has been looked at by researchers on how to improve the speed of search by reducing overhead costs. One of the ways to reduce the overhead cost is to incorporate parallelism to improve the performance of the BLAST algorithm. For this paper, I explored existing variations of parallel implementations of the BLAST algorithm and compared its performance improvements with that of serial implementation of BLAST. The speed-up efficiency noted by the parallel program is far greater compared to the serial program. The paper sheds light on the impact of parallelization of the BLAST algorithm and the advantages it has on the overall field of computational biology