%0 Journal Article
%T Features of Methylation and Gene Expression in the Promoter-Associated CpG Islands Using Human Methylome Data
%A Xin Du
%A Leng Han
%A An-Yuan Guo
%A Zhongming Zhao
%J International Journal of Genomics
%D 2012
%I Hindawi Publishing Corporation
%R 10.1155/2012/598987
%X CpG islands are typically located in the 5¡ä end of genes and considered as gene markers because they play important roles in gene regulation via epigenetic change. In this study, we compared the features of CpG islands identified by several major algorithms by setting the parameter cutoff values in order to obtain a similar number of CpG islands in a genome. This approach allows us to systematically compare the methylation and gene expression patterns in the identified CpG islands. We found that Takai and Jones¡¯ algorithm tends to identify longer CpG islands but with weaker CpG island features (e.g., lower GC content and lower ratio of the observed over expected CpGs) and higher methylation level. Conversely, the CpG clusters identified by Hackenberg et al.¡¯s algorithm using stringent criteria are shorter and have stronger features and lower methylation level. In addition, we used the genome-wide base-resolution methylation profile in two cell lines to show that genes with a lower methylation level at the promoter-associated CpG islands tend to express in more tissues and have stronger expression. Our results validated that the DNA methylation of promoter-associated CpG islands suppresses gene expression at the genome level. 1. Introduction CpG islands (CGIs), which are clusters of CpG dinucleotides in GC-rich regions, are often located in the 5¡ä end of genes and are considered as gene markers in vertebrate genomes [1¨C3]. These CpG islands, especially promoter-associated CpG islands, play important roles in gene silencing, genomic imprinting, X-chromosome inactivation, and tumorigenesis [4]. Due to the functional importance of CpG islands in transcriptional regulation and epigenetic modifications [5], multiple algorithms have been developed to identify CpG islands in a genome or a specific sequence. Overall, these algorithms can be classified into two groups: traditional algorithms and new algorithms. Traditional algorithms are based on three features and parameters (length, GC content, and ratio of the observed over the expected CpGs (CpG O/E)) [2, 4, 6, 7], while new algorithms are based on statistical property [8, 9]. Substantial debate exists as to which algorithm performs better and in which context, such as in organisms, tissues, or developmental stages [4, 8, 10¨C12]. Comparing different features of CpG islands, especially length of the predicted islands [11], our previous study suggested that Takai and Jones¡¯ algorithm is more appropriate overall for identifying promoter-associated islands of CpGs in vertebrate genomes [10]. However, the major
%U http://www.hindawi.com/journals/ijg/2012/598987/