I have a treatment group and a control group. I calculated the "contrast' between them using functions in Limma:

fit <- lmFit(array_data, design)
cont.matrix <- makeContrasts(Effect ="treat-control", levels = design)
fit2  <- contrasts.fit(fit, cont.matrix)
fit2  <- eBayes(fit2)

I am interested in calculating the effect size (Cohen's Q) and its variance of the treatment effect for each gene. The fit2 object does not explicitly contains those information. Any suggestion how to calculate the effect sizes and their variances from the fit2? In addition, how to find the degree of freedom for the t-statistics in fit2 object?

Thank you

Zicheng

It seems like Cohen's d would be more relevant here, in which case you can just follow the advice in:

A: Effect size similar to Cohen's d in limma

As for the degrees of freedom; if you poke around in the fit object, you would find some relevant fields:

fit$df.residual
fit$df.prior
fit$df.total

The total d.f. is that used in the moderated t-test - see Chapter 13 of the limma user's guide for more details.