A novel statistical measure for sequence comparison on the basis of k-word counts

Numerous efficient methods based on word counts for sequence analysis have been proposed to characterize DNA sequences to help in comparison, retrieval from the databases and reconstructing evolutionary relations. However, most of them seem unrelated to any intrinsic characteristics of DNA. In this paper, we proposed a novel statistical measure for sequence comparison on the basis of k-word counts. This new measure removed the influence of sequences’ lengths and uncovered bulk property of DNA sequences. The proposed measure was tested by similarity search and phylogenetic analysis. The experimental assessment demonstrated that our similarity measure was efficient.

► The increasing amount of gene sequences calls for efficient computational methods.
► Most of the methods view word frequencies as discrete units separately.
► We focus on correlations and bulk property of k-word.
► A new measure is proposed for sequence comparison on the basis of k-word counts.
► The experimental assessment demonstrated that our similarity measure is efficient.