To S+ group;
I am wording a slide for a presentation and I want to describe an a
posteriori power analysis of microarray data. There are 9 chips
experimental, 9 control. The average stdev of the ratios between control
and treatment is around 0.10 (pretty low), standardized for the magnitude of
the signal. I would like to clearly state (somehow) that one needs x number
of chips to detect a 1.3-fold change (delta) at alpha = 0.05, beta = 0.95.
The problem is that I want to explain a two-sided t-test using ratios
between control and treatment, not the actual values. My stdev is computed
from the log ratios. Is this an appropriate way to calculate sample size or
should I be using the actual values from control and treatment? Since my
stdev varies from gene to gene (there are essentially 20,000 separate
experiments per chip) I use the average stdev in order to calculate sample
size. Is this appropriate?
Thanks for any suggestions, I'm sure there's something wrong with this
approach but I can't find a paper describing anything similar.
Phillip Stafford
Amersham Genomics
_________________________________________________________________
Join the world?s largest e-mail service with MSN Hotmail.
http://www.hotmail.com
|