Study and Development of Novel Machine Learning Computational Framework for Identification of Fish Species using Molecular Markers

Abstract

newline The present work proposed to study and development of a novel machine learning computational frame system for identification of fish species using molecular marker. Species identification is the basic problem for the researcher in the scientific community. Advancement in computational technology and application of computer algorithm provided a platform for Machine Learning concept which improves with experiences. Machine Learning is responsible for computational development of analytical model for data analysis. Without any programming method computer can analyse the raw data with the help of iteration of the algorithm through Machine Learning technique. The key application of Machine Learning is analyzing of big data through its machine learning classifier. Over the last two decade, voluminous amount of biological raw data has been generated due to the advancement in technology and stored in a public repository database like NCBI. The challenging task for the researcher is how to extract information for application interest and processed of raw datasets from public domain. A novel DNAClassifier tool has been proposed developed which work on a pipeline of three algorithms. The first algorithm of DNAClassifier tool name Gene Parsing Algorithm generates curated datasets of nucleotide sequences from the downloaded raw nucleotide of molecular marker COI, Cytb, 12S rRNA and 16S rRNA. Second algorithm of DNAClassifier tool name Binary Conversion Algorithm takes input as first algorithm output and generates binary form of datasets after performance of 4 bit binary pattern feature extraction. Third algorithm of DNAClassifier tool name Train Test Data Algorithm takes input as second algorithm output and generates training and test datasets directly usable to any machine learning classifier based on ten folding method. After passing downloaded sequences of nucleotide molecular marker (mtDNA molecular marker) sequences through DNAClassifier tool we have 103326 numbers of records for COI mtDNA, 86889 numb