A Revamp Approach for Training of HMM to Accelerate Classification of 16S rRNA Gene Sequences

In the era of Information Technology, the field of Bioinformatics is rapidly growing with research in various related topics. The database of biological information has become much higher than its consumption. Automatic classification of biological information is one of the critical problems in Bioinformatics. Therefore, the critical issue is to regulate and manage the enormous amount of novel information to facilitate access to this useful and valuable biological information. The specific nucleus dilemma in classifying biological information is the annotation of various biological sequences with functional features. Annotation of the significant and rapidly increasing amount of genomic sequence data requires computational tools for classification of genes in DNA sequences. This paper presents a computational method for classification of highly conserved 16S rRNA biological sequences. We took Biological sequence classification as motivation to reveal a methodology that uses Hidden Markov Models (HMMs) to classify them. This paper explains the description of the algorithms used for implementing three phases of HMM (training, decoding, and evaluation) to classify sequences into clusters that have known similar functional properties. In the implementation of the training phase, we have addressed practical issues like initial parameter selection for HMM and computational weakness for the large data set. Later in the paper, we have shown that methodology presents a classification accuracy of 91% for Bacillus and 97% for Clostridia.