Prediction of Eukaryotic Exons via the Singularity Detection Algorithm
The prediction of eukaryotic exons is an important topic in bioinformatics. In this paper, a model-independent
method based on the singularity detection (SD) algorithm and the three-base periodicity has been developed for predicting
exons in DNA sequences of eukaryotes. Using the HMR195 data set, BG570 data set and 200 test data as test sets, we
show that, (1) In comparison with the exon prediction by nucleotide distribution (EPND), modified Gabor-wavelet
transform (MGWT) and fast Fourier transform plus empirical mode decomposition (FFTEMD) method, the proposed SD
method notably improves prediction accuracy of exons, especially short exons or the ability to discern two contiguous
short exons disunited by a short intron; (2) The SD method also significantly enhances the performance of the noise
suppression in exon prediction over all assessed model-independent methods. The performance of the SD method is
evaluated in terms of the signal-to-noise, the approximate correlation, the area under the receiver operating characteristic
curve and the accuracy against those of the EPND, MGWT and FFTEMD method over HMR195 data set, BG570 data set
and 200 test data. Experimental results demonstrate that the SD method outperforms all assessed model-independent
methods with respect to those performance parameters.
Keywords: Bioinformatics, eukaryote, exon, model-independent method, singularity detection, three-base periodicity.
Rights & PermissionsPrintExport