Bootstrap Testing and Inference for Recurrent DNA Copy Number Aberrations

• Activity Number: 242
• Type: Contributed
• Date/Time: Monday, July 30, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Vonn Walter*+ and Fred Wright and Andrew B Nobel and Matthew D. Wilkerson and David Neil Hayes
• Companies: The University of North Carolina at Chapel Hill and The University of North Carolina at Chapel Hill and The University of North Carolina at Chapel Hill and The University of North Carolina at Chapel Hill and The University of North Carolina at Chapel Hill
• Keywords: DNA copy number ; Cancer ; Bootstrap ; Microarray

Abstract

DNA copy number aberrations (CNAs) occur when at least one copy of a chromosomal region is gained (amplification) or lost (deletion). Genomic instability can lead to CNAs at random locations throughout the genome. However, some CNAs, termed recurrent, can be found in the same genomic region in multiple samples. In tumor tissue, recurrent CNAs may provide a selective growth advantage because they affect genes associated with tumor genesis and proliferation. Most methods for detecting recurrent CNAs employ testing procedures that compute local summary statistics and use permutation-based null distributions to assess statistical significance. However, there are reasons why bootstrap resampling provides an attractive alternative to permutation. We present bootstrap methods for assessing the significance of recurrent CNAs, as well as the significance of the association between copy number (CN) values at a given locus and covariate data. We also introduce bootstrap methods for computing confidence intervals for recurrent CNAs and confidence intervals for CN loci associated with covariate data. Applying these methods in real tumor datasets produces biologically relevant results.