Detection of Local Signals in Genomics

• Activity Number: 36
• Type: Contributed
• Date/Time: Sunday, August 4, 2013 : 2:00 PM to 3:50 PM
• Sponsor: IMS
• Author(s): David Siegmund*+ and Benjamin Yakir Yakir and Nancy Zhang
• Companies: Stanford University and Hebrew University of Jerusalem and University of Pennsylvania
• Keywords: change-point ; point process ; read depth ; copy number

Abstract

Several problems of genomic analysis involve detection of local genomic signals. When the data are generated by sequence based methods, the variability of read depth at different positions on the genome suggests point process models involving non-homogeneous Poisson processes, or perhaps negative binomial processes to accommodate excess variability. We discuss a number of examples, including copy number variation inferred from read depth and/or paired end reads, and detection of protein-DNA interactions by ChIP-Seq. Approaches to these problems based on change-point methods are described, with particular attention to the problems of multiple comparisons. The theoretical results are illustrated on simulated and on real data.