Adaptive exon prediction using maximum modified normalized algorithms

Abstract

Exact identification of exon fragments in a deoxyribonucleic acid (DNA) sequence is a critical task in the field of genomics. This is a crucial part in finding health disorders and design drugs. Exons are the infoessentialin coding of proteins in DNA. HenceforwardfindingsuchDNA sectionsremains important part of genomics. In DNA arrangement, nucleotides form the key elementary units. Three base periodicity (TBP) is a basic property displayed by only exon fragments, and is not shown in other DNA sections that could beforecastedeasily with techniques of signal processing. Frommanymethods, adaptive methodswerefavorablebecause of theircompetencein altering weight coefficients depending on gene sequence. Hence, an adaptive exon predictor (AEP) is proposedwithMaximum Modified Normalized Least Mean Square (MMNLMS) algorithm. TheAEP derived using MMNLMS is combined with its sign versions to decrease complexity in computations. Also, this was clear thatModified Normalized Sign Regressor LMS (MMNSRLMS) based AEPwas more effective in exon identification applications withmetricsalikeSpecificity, Sensitivity, and Precision. Thus, computational complexity is greatly minimized, and AEPs proposedweresuitablefor use in nano devices. Lastly exon findingcapabilitywithdiverse AEPs stands verified with DNA datasets from National Center for Biotechnology Information (NCBI) gene databank.