dPeak: A Statistical Framework for Identifying Closely Spaced Protein-DNA Interaction Events Using ChIP-Seq Data

• Activity Number: 248
• Type: Contributed
• Date/Time: Monday, July 30, 2012 : 2:00 PM to 3:50 PM
• Sponsor: ENAR
• Author(s): Dongjun Chung*+ and Sunduz Keles
• Companies: University of Wisconsin-Madison and University of Wisconsin-Madison
• Keywords: ChIP-Seq ; Next generation sequencing ; Deconvolution ; Paired-end tag ; PET

Abstract

ChIP-Seq is a powerful approach for genome-wide profiling of transcription factor binding sites, histone modifications, and nucleosome occupancy. Identification of binding sites using ChIP-Seq data is especially challenging in studies of prokaryotic genomes, such as Escherichia coli, because it requires distinction of closely spaced binding sites, some of which are separated by only few base pairs. To address this research question, we developed dPeak, a high resolution binding site identification (deconvolution) algorithm using ChIP-Seq data. The dPeak algorithm implements a probabilistic model that accurately describes ChIP-Seq data generation process. We showed that by utilizing this algorithm, closely spaced binding sites can be individually identified with high accuracy, especially when paired-end tag (PET) data is utilized. Theoretical study of this probabilistic model elucidated when and why prediction based on single-end tag (SET) data can be degraded.