期刊
BIOINFORMATICS
卷 32, 期 12, 页码 1766-1772出版社
OXFORD UNIV PRESS
DOI: 10.1093/bioinformatics/btw058
关键词
-
类别
资金
- Direct Grant of the Chinese University of Hong Kong [CUHK4053139]
Motivation: Understanding the interactions of different DNA binding proteins is a crucial first step toward deciphering gene regulatory mechanism. With advances of high-throughput sequencing technology such as ChIP-seq, the genome-wide binding sites of many proteins have been profiled under different biological contexts. It is of great interest to quantify the spatial correlations of the binding sites, such as their overlaps, to provide information for the interactions of proteins. Analyses of the overlapping patterns of binding sites have been widely performed, mostly based on ad hoc methods. Due to the heterogeneity and the tremendous size of the genome, such methods often lead to biased even erroneous results. Results: In this work, we discover a Simpson's paradox phenomenon in assessing the genome-wide spatial correlation of protein binding sites. Leveraging information from publicly available data, we propose a testing procedure for evaluating the significance of overlapping from a pair of proteins, which accounts for background artifacts and genome heterogeneity. Real data analyses demonstrate that the proposed method provide more biologically meaningful results. Availability and implementation: An R package is available at http://www.sta.cuhk.edu.hk/YWei/ChIPCor.html.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据