ArrayPipe Online Documentation
[← previous] [up ↑] [next →]
Limma's empirical Bayes moderated t-test
Through Bioconductor's limma package ArrayPipe offers a new statistical test from version 2.0 onwards, the empirical Bayes moderated t-statistic:
From the limma usersguide:
"For statistical analysis and assessing differential expression, limma uses an empirical Bayes
method to moderate the standard errors of the estimated log-fold changes. This results in
more stable inference and improved power, especially for experiments with small numbers
of arrays."
(for more info see limma package site).
The module also incorporates correction for multiple testing, but this is
turned off by default.
In the settings section you can chose amongst methods such as 'holm',
'hochberg', 'hommel', 'bonferroni', 'BH' (Benjamini-Hochberg = fdr), and 'BY' (Benjamini-Yekutieli):
Warning: By default, the moderated t-test assigns p-values for spots that lack replicates (when, for example, the other instances have been flagged out). It's basing this on the standard variation of other probes (see statement from the limma author on this). If you don't feel comfortable with this you can prevent 'single' values to get p-values by checking the according box in the settings:
Note: The 'include dye-effect' option only has an effect in unbalanced designs, i.e. where the numbers of forward and reverse labelled slides are not equal.
Home
last modified $Date: 2007/08/27 14:02:03 $
for questions or remarks e-mail karsten_hokamp@sfu.ca.