Parameters Estimation of HMM Model for the Classification of Biological Sequences

Shitalkumar R. Sukhdeve, Manish P. Kurhekar

Abstract: Classification of the biological data is a fundamental topic of research in computer science, especially in bioinformatics. Given an unknown sequence, finding the common patterns in it which appears in the known sequences and placing it into the most probable family is the classification problem. The bacteria?s of bacillus and clostridia are so conserved that they show nearly same characteristics which make them difficult to classify. In this paper we have tried to classify such biological sequences with the help of Hidden markov model (HMM) in combination with the Baum Welch model. In this experiment, we have tried to estimate the parameters of HMM so that the classification process should converge in the minimum number of iterations. We have achieved approximately 89% and 90.73% accuracy of classification in case of bacillus and clostridia respectively.

Keywords: Hidden Markov Model, HMM, DNA Sequences, biological, Baum Welch algorithm, parameter estimation