CNV-Guided Multi-Read Allocation for ChIP-Seq

• Activity Number: 453
• Type: Contributed
• Date/Time: Wednesday, August 6, 2014 : 8:30 AM to 10:20 AM
• Sponsor: WNAR
• Author(s): Qi Zhang*+
• Companies: University of Wisconsin-Madison

Abstract

In ChIP-seq and other short read sequencing experiments, a considerable amount of reads can be aligned to multiple locations on the reference genome (multi-reads). Allocating the multi-reads to appropriate positions are important for the downstream analysis, especially when the repetitive regions are concerned. The state-of-the-art multi-read allocation algorithms rely on the uni-read content around the alignment locations. In the highly repetitive regions where such content is low, the existing algorithms may not perform reliably, and we need external information for help. Copy-number variation (CNV) is known to have a direct impact on the read densities, and has the potential of improving multi-read allocation.

We propose cnvCSEM (cnv-guided ChIP-Seq by Expectation-Maximization algorithm), a flexible framework that incorporates CNV in multi-read allocation. cnvCSEM initializes the EM algorithm with CNV-aware initial values. Our biology motivated simulations shows that cnvCSEM enjoys higher coverage in read allocation and higher accuracy in read-depth recovery. The biological relevance of the allocated reads of cnvCSEM and the resultant peaks are also evaluated in the analysis of