Improving the Effectiveness of BLAST Search Using Genetic Algorithms

Abstract

The BLAST (Basic Local Alignment Search Tool) is for nding similarity between sequences by using local alignment. This tool compares nucleotide or protein sequences with sequence databases and calculates the statistical signi cance of the matches. Set of parameters are used in similarity search and they are assigned with default values initially. But default parameter values are not always suitable for nding sequence similarity. Parameter values depend on the type of the similarity search. Signi cant results can be obtained by changing the parameters appropriately. This study is an e ort to nd the best parameter settings for a given nucleotide query sequence. We optimized four parameters of BLAST tool by using genetic algorithms and identi ed that several parameter settings that achieve better results than with default parameter settings. When changing the length of the sequence we identi ed default parameters set was not always good.