Probabilistic Measures of Similarity/Dissimilarity Between Markov Models for Construction of Guide Tree for Multiple Sequence Alignment

In this paper, we explore the consequences of changing the method of construction of guide tree used for the progressive alignment. The method is based on the concept of comparing the similarity/dissimilarity between two Markov models using Kullback–Leibler divergence for construction of pair-wise distance matrix. We evaluated both the leading MSA method, ClustalW as well as the new MSA method which we have developed on benchmark DNA datasets. Improvement in alignment accuracy (i.e. sum of pair’s scores) is observed when compared with ClustalW method.