Model-Based Stastical Approach for Paired-End ChIP-Seq

• Activity Number: 253
• Type: Contributed
• Date/Time: Monday, July 30, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Sangsoon Woo*+ and Xuekui Zhang and Raphael Gottardo
• Companies: Fred Hutchinson Cancer Research Center and The Johns Hopkins University and Fred Hutchinson Cancer Research Center
• Keywords: ChIP-Seq ; Paired-End sequencing ; PING

Abstract

ChIP-Seq enables to study protein interaction with DNA such as histones,histone modification,nucleosome positioning and transcription binding sites. Several statistical approaches have been developed for single-ended(SE) ChiP-Seq to identify genomic regions showing biological signal. The paired-end(PE) sequencing platform is recently introduced technology, which generates double-end sequencing reads. Compared to SE, PE offers better information of each DNA fragment by providing genome coordinates of both ends. We recently published model-based approach for identifying nucleosome positions(PING) that models bi-directional read densities using mixture models and imputing missing reads. As a modification of PING, we developed a model-based approach for analyzing PE data(PING-PE) by forcing fragment's forward and reverse reads belong to the same mixture component, modifying parameter estimates, imputing missing reads and changing segmentation step. We will discuss about efficiency of PE sequencing compared to SE sequencing and also compare our PING-PE approach to several widely used ChIP-Seq data analysis tools such as MACS, CisGenome and SIPeS using published PE sequencing data.