Clustering with Exclusion Zones: Genomic Applications

• Activity Number: 581
• Type: Contributed
• Date/Time: Wednesday, August 4, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Mark Segal*+
• Companies: University of California, San Francisco
• Keywords: Clump ; Chromosome ; Scan Statistic ; Repeat Sequence

Abstract

Methods for formally evaluating the clustering of events in space or time, notably the scan statistic (SS), have been widely applied. In order to utilize the SS it is necessary to know the extent of the domains wherein the events arise. It is implicit that these domains have no exclusion zones (EZs) in which events cannot occur. However, this requirement is often not met. When EZs are known it is easy to correct the SS by simply adjusting the domain extent. Here, we tackle the more ambitious goal of evaluating clustering when EZs are unknown. By examining the behavior of clumps over all cluster counts and lengths, we show that the EZs impart a characteristic signature. We devise an algorithm to estimate EZ extent based on this. Performance is assessed via simulation. Genomic applications to binding sites, housekeeping genes, and microRNAs reveal dramatic inferential changes.