Title: A novel approach to reduce computational load in least norm super resolution gene predictor

Authors: Manidipa Roy; Soma Barman

Addresses: Institute of Radio Physics & Electronics, University of Calcutta, Kolkata 700009, India ' Institute of Radio Physics & Electronics, University of Calcutta, Kolkata 700009, India

Abstract: Recent techniques of spectrum estimation focus on linear algebraic concept of subspaces. Spectral estimation based on noise subspace method has already been used for DNA sequence analysis having significance in genomic study. It is known that exons show 3-base periodicity property, which is used to identify protein coding regions in gene. In an earlier work, the authors developed the least norm estimator showing sharp period-3 peaks in coding regions completely eliminating background noise. But limitation of the method was its high computational load. In this work a technique to reduce computation time is introduced enhancing the computational efficiency to a great extent. Moreover statistical classification criteria termed as AIC (Akaike Information Criterion) has been incorporated to resolve the problem of estimation of the signal and noise subspace dimension. Receiver Operating Characteristics (ROC) analysis, a quantitative measurement metric is used to evaluate the least norm and modified periodogram predictors.

Keywords: periodograms; deoxyribo nucleic acid; DNA sequencing; least norm solution; eigenvector; eigenvalue; autocorrelation; computational load; super-resolution gene predictor; spectrum estimation; bioinformatics; protein coding regions; receiver operating characteristic; ROC analysis.

DOI: 10.1504/IJDMB.2016.080672

International Journal of Data Mining and Bioinformatics, 2016 Vol.16 No.3, pp.230 - 251

Accepted: 11 Sep 2016
Published online: 01 Dec 2016 *

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