Novel exon predictors using variable step size adaptive algorithms

Abstract

Regions which code for protein in deoxyribonucleic acid (DNA) sequence are crucial to determine in bioinformatics field. Exon region study is significant in drug design also disease detection. Three basic periodicity (TBP) have been noted in exons. Adaptive techniques of signal processing were likely with unique capability to alter genome sequence dependent weight coefficients. From these, we present a new adaptive exon predictor (AEP) using variable step size least mean square (VSLMS) algorithm along with its signed versions that includes SRVSLMS, SVSLMS also SSVSLMS algorithms to decrease computational complexity and compared to LMS. SRVSLMS based AEP was shown to be better based on metrics like Precision 0.6751, Sensitivity 0.6749, also Specificity 0.6625 at threshold value 0.8. Thus, numerous AEPs are able to predict exon positions with distinct real genomic sequences of Homo sapiens as of National Centre for Biotechnology Information (NCBI) gene databank.