Finding Exact and Solo LTR-Retrotransposons in Biological Sequences Using SVM

Finding repetitive subsequences in genome is a challengeable problem in bioinformatics research area. A lot of approaches have been proposed to solve the problem, which could be divided to library base and de novo methods. The library base methods use predetermined repetitive genome’s subsequences, where library-less methods attempt to discover repetitive subsequences by analytical approaches. In this article we propose novel de novo methodology which stands on theory of pattern recognition’s science. Our methodology by using Support Vector Machine (SVM) classification and clustering methods could extract exact and Solo LTR-retrotransposons. This methodology issued to show complexity efficiency and applicability of the pattern recognition theories in bioinformatics and biomathematics research areas. We demonstrate applicability of our methodology by comparing its results with other well-known de novo method. Both applications return classes of discovered repetitive subsequences, were their results when had applied on show more that 90 percents similarities.