Nucleosome positioning plays an important role in predicting the methylation status of CpG islands

Abstract

CpG island methylation is highly correlated with epigenetic gene control during mammalian development. The majority of CpG islands are normally unmethylated, but in some specifically pathological situations some of the CpG islands are prone to become methylated. Current methods about prediction of DNA methylation utilized DNA sequence features, transcription factor binding site features and histone methylation mark features. In this study, we used SVM to predict the methylation status of CpG islands from human Chromosome 22 and improved the accuracy to 90.6%. We not only used 4 DNA sequence features such as sequence length, CG ratio, C+G% content, CpG frequency and 39 transcription factor binding sites features, but also added nucleosome position features. Our results support the view that nucleosomes are preferentially targeted by DNA methyltransferases and imply that nucleosome-bound DNA regions are more prone to become methylated than flanking regions. At last, we explain why the histone methylation marks can predict DNA methylation status.

Related Papers

• → Genome-wide analysis of aberrant methylation in human breast cancer cells using methyl-DNA immunoprecipitation combined with high-throughput sequencing(2010)196 cited
• → Genome-wide analysis of DNA methylation in bovine placentas(2014)94 cited
• → DNA Methylation Profiling in Zebrafish(2011)33 cited
• → A Human DNA Methylation Site Predictor Based on SVM(2009)2 cited
• → Sodium Bisulfite Conversion of Human Genome for DNA Methylation Studies(2020)