Multiple contrasts

[Arthfael]

Dear team,
When multiple contrasts are to be tested, limma recommends to run all in one call because the variance estimates then benefit from the extra data. For the t-test I will get one. p-value per contrast. As I understand DEqMS is meant as an extension of limma's moderated t-test, so similarly I would expect as many p-values as I have contrasts. Yes I just ran it on a fit with 2 contrasts and got only one sca-P-value column. Am I doing something wrong, or are so far only single contrasts supported?

[Arthfael]

Nevermind, that's what the coef_col argument in outputResult() if for...

[Arthfael]

I did notice though that the P-values outputResult() outputs (not the sca-P-values, the original ones in column P.Value) are different from the limma ones. Are these normal (not moderated, e.g. Welch or Student) t-test P values?

[yafeng]

@Arthfael the column P.Value are Limma reported results. If you get different values, then check if the input gene expression matrix and design matrix used in fit1 is the same or not.

Yafeng