|
Activity Number:
|
406
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Wednesday, August 5, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #304394 |
|
Title:
|
Detecting Breakpoints Using Multiscale Wavelet Products
|
|
Author(s):
|
Li Hsu*+
|
|
Companies:
|
Fred Hutchinson Cancer Research Center
|
|
Address:
|
1100 Fairview Ave. N., M2-B500, Seattle, WA, 98109-1024,
|
|
Keywords:
|
wavelet ; multi-scale product ; breakpoint
|
|
Abstract:
|
Copy number changes occur in many diseases including various cancers and genetic disorders. Recent technologic developments allow researchers to examine genome-wide copy number changes at thousands of loci simultaneously. We proposed a wavelet-based nonparametric approach for detecting the break points where copy number changes occur. The maximum of two-scale wavelet products across scales, as a novel test statistic, was taken for combining information across scales to improve power. To avoid the false positives caused by the local spatial correlation among adjacent wavelet coefficients, we proposed an approach which examines only the locations at which local maxima occur. Statistical inference was provided for the proposed statistic and the finite sample performance was examined by both simulated and real data sets.
|
