An Improved K-Means Algorithm for DNA Sequence Clustering

Abstract

In recent years, billions of DNA and protein sequences are subject to sequencing. However, few of them have known structures and functions, most remain unknown. The solution to this problem is to link sequences between them rather than revisit each new sequence independently of other sequences. Thus, if we manage to assimilate a sequence S1 to another sequence S2 or to a group of previously studied sequences, this will allow us to directly deduce the structure, functions and phylogenetic classification of S2. The purpose of this work is to adapt clustering methods to the specific problem of classification of DNA sequences. We introduce a new method based on K-means clustering for DNA sequences clustering. We begin by explaining and motivating our approach, then we present obtained results.