Get access

Detecting Genomic Aberrations Using Products in a Multiscale Analysis

Authors

  • Xuesong Yu,

    Corresponding author
    1. Statistical Center for HIV/AIDS Research and Prevention, Fred Hutchinson Cancer Research Center, 1100 Fairview Avenue N., Seattle, Washington 98109, U.S.A.
    Search for more papers by this author
  • Timothy W. Randolph,

    Corresponding author
    1. Biostatistics and Biomathematics Program, Fred Hutchinson Cancer Research Center, 1100 Fairview Avenue N., Seattle, Washington 98109, U.S.A.
    Search for more papers by this author
  • Hua Tang,

    Corresponding author
    1. Department of Genetics, Stanford University School of Medicine, Stanford, California 94305, U.S.A.
    Search for more papers by this author
  • Li Hsu

    Corresponding author
    1. Biostatistics and Biomathematics Program, Fred Hutchinson Cancer Research Center, 1100 Fairview Avenue N., Seattle, Washington 98109, U.S.A.
    Search for more papers by this author

email:xyu@fhcrc.org

email:trandolp@fhcrc.org

email:huatang@stanford.edu

email:lih@fhcrc.org

Abstract

Summary Genomic instability, such as copy-number losses and gains, occurs in many genetic diseases. Recent technology developments enable researchers to measure copy numbers at tens of thousands of markers simultaneously. In this article, we propose a nonparametric approach for detecting the locations of copy-number changes and provide a measure of significance for each change point. The proposed test is based on seeking scale-based changes in the sequence of copy numbers, which is ordered by the marker locations along the chromosome. The method leads to a natural way to estimate the null distribution for the test of a change point and adjusted p-values for the significance of a change point using a step-down maxT permutation algorithm to control the family-wise error rate. A simulation study investigates the finite sample performance of the proposed method and compares it with a more standard sequential testing method. The method is illustrated using two real data sets.

Ancillary