Application of clustering technique in multiple sequence alignment

Abstract

This article presents a new approach using clustering technique for creating multiple sequence alignments. Currently, the most widely used strategy is the progressive alignment. However, each step of this strategy might generate an error which will be low for closely related sequences but will increase as sequences diverge. For that reason, determining the order in which the sequences will be aligned is very important. Following this idea, we propose the application of a clustering technique as an alternative way to determine this order. To assess the reliability of this new strategy, two methods were modified in order to apply a clustering technique. The accuracy of their new versions was tested using a reference alignment collection. Besides, the modified methods were also compared with their original versions, obtaining better alignments. © Springer-Verlag Berlin Heidelberg 2005.