|
Activity Number:
|
194
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 4, 2008 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
IMS
|
| Abstract - #301945 |
|
Title:
|
Estimating Individual Effect Sizes and Local FDR in Multiple Hypothesis Testing
|
|
Author(s):
|
József Bukszár*+ and Edwin J.C.G. van den Oord
|
|
Companies:
|
Virginia Commonwealth University and Virginia Commonwealth University
|
|
Address:
|
410 North 12th Street, Medical College of Virginia Campus, Richmond, VA, 23298,
|
|
Keywords:
|
multiple hypothesis testing ; individual effect size ; local FDR ; genome-wide association
|
|
Abstract:
|
We present a method that estimates the individual effect sizes (IES) in multiple-hypothesis testing problem. A novel and key element is that we use the real likelihood function of the multiple tests rather the one based on the frequently used mixture model. Interestingly enough, this latter one would result in fairly inaccurate estimate even for arbitrarily high sample size. Our IES estimates enable us to estimate the posterior probability that a hypothesis is true null, i.e. the local FDR, accurately. Our methods will be demonstrated on analysis of genome-wide association data.
|
