I just wanted to alert the community to an issue I discovered while using limma on the Galaxy webserver.
The issue arises when conducting differential expression between contrasts. The process is described on the galaxy server as follows:
"Names of two groups to compare separated by a hyphen e.g. Mut-WT. If the order is Mut-WT the fold changes in the results will be up/down in Mut relative to WT. If you have more than one contrast enter each separately using the Insert Contrast button below. For differences between contrasts use e.g. (Mut1-Ctrl1)-(Mut2-Ctrl2). For more info, see Chapter 8 in the limma User's guide: https://www.bioconductor.org/packages/release/bioc/vignettes/limma/inst/doc/usersguide.pdf"
The issues is that unlike "Mut-WT" (where the fold changes in the results will be up/down in Mut relative to WT as described in the galaxy documentation), when one uses "(Mut1-Ctrl1)-(Mut2-Ctrl2)" the fold changes in the results are instead up/down in (Mut2-Ctrl2) relative to (Mut1-Ctrl1) . I'm not sure if this is an error with the Galaxy implementation of limma or if this is something more fundamental.
Also please note that would not be an issue if the file or output indicated which sample/contrast is used as the numerator or denominator, however neither the file name nor the file itself indicated which is which. You only have a column that reads "logFC" and if the user is not careful they will end up with completely inverted data.
Anyway - hope this saves someone some headache/time.
