|
Activity Number:
|
274
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 10, 2004 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #300918 |
|
Title:
|
FDR-based Methods for Determining Replicate Size For Microarray Experiments
|
|
Author(s):
|
Vishwanath Iyer*+ and Sanat K. Sarkar
|
|
Companies:
|
Bristol-Myers Squibb Company and Temple University
|
|
Address:
|
5 Research Pkwy., Wallingford, CT, 06492,
|
|
Keywords:
|
replicate size ; FDR ; microarray ; FNR ; sample size
|
|
Abstract:
|
One issue that has received considerable attention recently has been the determination of the number of DNA array replicates that one needs to use to be "reasonably certain" that the differences one observes between control and treatments are "real." The challenge has been to strike a balance between the cost of expensive replicates and some measure of statistical judiciousness. The exploratory nature of these experiments allows us the leeway of considering methods based on the less strict criterion of controlling the False Discovery Rate (FDR) instead of the more conservative Familywise Error Rate (FWER). Previously, we derived an exact formula for FDR and its conjugate quantity, the False Negative Rate (FNR) and described a method to select a replicate size based on these two criteria under certain stringent assumptions. We compared different multiplicity adjustment methods, and their effect on replicate size. We propose an approximation to the exact formula that significantly decreases computation time. We explore relaxing the normality assumptions from the previous paper.
|
