Title: A classifier system for predicting RNA secondary structure

Authors: Monther Aldwairi; Bashar Al-Hajasad; Yaser Khamayseh

Addresses: Faculty of Computer and Information Technology, Jordan University of Science and Technology, Irbid 22110, Jordan ' Faculty of Computer and Information Technology, Jordan University of Science and Technology, Irbid 22110, Jordan ' Faculty of Computer and Information Technology, Jordan University of Science and Technology, Irbid 22110, Jordan

Abstract: Finding the secondary structures of ribonucleic acid sequences is a very important task. The secondary structure helps determine their functionalities which in turn plays a role in the proteins production. Manual laboratory methods use X-ray diffraction to predict secondary structures but it is complex, slow and expensive. Therefore, different computational approaches are used to predict RNA secondary structure in order to reduce the time and cost associated with the manual process. We propose a system called IsRNA to predict a single element, internal loop, of the RNA secondary structure. IsRNA experiments with different classifiers such as SVM, KNN, Naive Bayes and Simple K means to find the most accurate classifier. We present a through experimental evaluation of 24 features, classified into five groups, to determine the most relevant feature groups. The system is evaluated using Rfam sequences and achieves an overall sensitivity, selectivity, and accuracy of 96.1%, 98%, and 96.1%, respectively.

Keywords: bioinformatics; RNA secondary structure; internal loop; classifiers; ribonucleic acid; RNA sequences; protein production; SVM; support vector machine; KNN; k-nearest neighbour; naive Bayes; simple K-means.

DOI: 10.1504/IJBRA.2014.060764

International Journal of Bioinformatics Research and Applications, 2014 Vol.10 No.3, pp.307 - 320

Received: 21 Sep 2011
Accepted: 30 Jul 2012
Published online: 24 Oct 2014 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article