Prediction of CpG island methylation status by integrating DNA physicochemical properties

Abstract

As an inheritable epigenetic modification, DNA methylation plays important roles in many biological processes. The non-uniform distribution of DNA methylation across the genome implies that characterizing genome-wide DNA methylation patterns is necessary to better understand the regulatory mechanisms of DNA methylation. Although a series of experimental technologies have been proposed, they are cost-ineffective for DNA methylation status detection. As complements to experimental techniques, computational methods will facilitate the identification of DNA methylation status. In the present study, we proposed a Naïve Bayes model to predict CpG island methylation status. In this model, DNA sequences are formulated by “pseudo trinucleotide composition” into which three DNA physicochemical properties were incorporated. It was observed by the jack-knife test that the overall success rate achieved by the proposed model in predicting the DNA methylation status was 88.22%. This result indicates that the proposed model is a useful tool for DNA methylation status prediction.