A new method for estimating parameters of a profile hidden Markov model based on phylogenetic tree

A new method for estimating parameters of a profile hidden Markov model based on phylogenetic tree

Hidden Markov Models (HMM) are popular statistical tool for modelling a wide range of time series data. These Models are practical tools which have been successfully implemented in many fields of bioinformatics. They are applied to protein sequence alignment, protein family annotation and gene-finding. A Profile Hidden Markov Model (PHMM) is a standard form of an HMM representing a class of Left-to-Right model used for modelling protein and DNA sequence families based on multiple alignment. It is possible to train the PHMM di y from unaligned sequences. The BaumWelch algorithm and the Bayesian Monte Carlo Markov Chain (BMCMC) method are well known approaches for estimating parameters of HMMs. In this paper, we first implement two methods for estimating parameters of small artificial PHMM with 6 Match states and mid entropy. Secondly, in order to improve the prediction accuracy of the estimation of the parameters of the PHMM. we classify the training data based on their phylogenetic tree. We finally apply an heuristic algorithm for estimating parameters of the PHMAI. It is expected that the use of our methodology improves the precision of parameter estimation considerably. All the program is available at http://www.bioinf.cs.ipm.ir/softwares/CSPHTREE/.