Proper

Abstract

MicroRNA (miRNA) analysis and classification research have progressed rapidly in last decade.Various tools and techniques were developed for miRNA identification using various machine learning techniques. Very little work has been done in exploring dominating attributes. Dominating attributes will reduce the complexity of the model and classification accuracy will also be enhanced.. We have studied miRNA sequences from Oryza , Zea mays, Sorghum, Arabidopsis and Brassicaand derived 14 attributes for precursors, 9 attributes for mature miRNAs and 20 attributes combining the both. These attributes were used for relevance analysis. Principal component analysis and information gain techniques were used for dominating attributes prediction. In our study we derived that decision trees based classifications based on J48 algorithm are best suited for the analytical study of miRNA. The derived dominating attributes are biologically significant as evidences from wet lab experiments. Decision trees based classifications of plant miRNA and the derived dominating attributes through machine learning approach are matching with the lab results.