Exploring the Cooccurrence Patterns of Multiple Sets of Genomic Intervals

• Type of Material: Article
• Authors: Hao Wu, Emory University; Zhaohui Qin, Emory University
• Language: English
• Date: 2013-05-04
• Publisher: Hindawi Publishing Corporation
• Publication Version: Final Published Version
• Copyright Statement: © 2013 Hao Wu and Zhaohui S. Qin.
• License: Creative Commons Attribution 3.0 Unported
• Final Published Version (URL): http://www.hindawi.com/journals/bmri/2013/617545/
• Title of Journal or Parent Work: BioMed Research International
• ISSN: 2314-6133
• Volume: 2013
• Issue: 2013
• Start Page: 1
• End Page: 7
• Grant/Funding Information: Hao Wu is partially funded by PHS Grant UL1 R025008 from the Clinical and Translational Science Award program, National Institute of Health, National Center for Research Resources.; Zhaohui S. Qin is partially funded by NIH Grant R01HG005119.
• Research Categories: Biology, Biostatistics

Abstract

Background. Exploring the spatial relationship of different genomic features has been of great interest since the early days of genomic research. The relationship sometimes provides useful information for understanding certain biological processes. Recent advances in high-throughput technologies such as ChIP-seq produce large amount of data in the form of genomic intervals. Most of the existing methods for assessing spatial relationships among the intervals are designed for pairwise comparison and cannot be easily scaled up. Results. We present a statistical method and software tool to characterize the cooccurrence patterns of multiple sets of genomic intervals. The occurrences of genomic intervals are described by a simple finite mixture model, where each component represents a distinct cooccurrence pattern. The model parameters are estimated via an EM algorithm and can be viewed as sufficient statistics of the cooccurrence patterns. Simulation and real data results show that the model can accurately capture the patterns and provide biologically meaningful results. The method is implemented in a freely available R package giClust. Conclusions. The method and the software provide a convenient way for biologists to explore the cooccurrence patterns among a relatively large number of sets of genomic intervals.