Analysis of codon bias pattern for genome sequences through a three-phase network model

Abstract

Codon bias generates specific patterns of synonymous codons usage in genomes of all texta. Codon bias patterns of genomes are extracted here in spectral domain and named as Codon Bias Spectra (CBS). Such spectral models of codon bias are demonstrated through a three-phase network model of various genomes. GC, GC3 and AT3 bias patterns along with codon spectra are shown in the paper by modifying the parameters in this spectral model. Patterns of CBS are studied here for codon bias in prokaryotic genomes. Correlation and regression analysis of CBS show the influencing effects of GC, GC3 and AT3 bias. Analysis of the area under ROC curve (AUC) is performed for testing the statistical significance of the three-phase network model for identifying codon bias locations in genome.