Locating CpG Islands with Kullback-Leibler DivergenceYung-Pin Chen*, Andrew Dittmore, Yasuhiro Goda, Alicia Laughton and Jessica Minnier
Department of Mathematical Sciences, Lewis & Clark College, Portland, USA
- *Corresponding Author:
- Yung-Pin Chen
Department of Mathematical Sciences
Lewis & Clark College, Portland, USA
E-mail: [email protected]
Received date: May 24, 2012; Accepted date: June 21, 2012; Published date: June 23, 2012
Citation:Chen YP, Dittmore A, Goda Y, Laughton A, Minnier J (2012) Locating CpG Islands with Kullback-Leibler Divergence. J Biomet Biostat 3:148 doi:10.4172/2155-6180.1000148
Copyright: © 2012 Chen YP, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
A CpG island is a short contiguous DNA subsequence that is rich in CG dinucleotides. CpG islands are often located around the promoters of housekeeping genes and have been found associated with certain tissue-specific genes. This observation indicates that they can be used as markers to identify genes. The information about the locations of CpG islands can also help us understand a gene regulation process called methylation. In this report, we propose a statistical method for locating CpG islands. Our method employs the Kullback-Leibler divergence. We use the given DNA sequence to determine a window size and a shift size for computing the divergence values along a DNA segment. A region in the proximity of a CpG island should contain consecutive windows with high divergence values. The distribution of the Kullback-Leibler divergence values can be suitably fitted by a truncated Pareto distribution. We estimate the parameters of the truncated Pareto distribution via the maximum likelihood principle. Then the fitted distribution is applied to locate regions with a divergence value exceeding a threshold level of significance. To assess the accuracy of our method, we compare our results to the putative CpG islands found in four well-studied mouse and human DNA sequences. The comparison suggests our approach consistently yields reliable predictions of CpG island locations.