Hello,

I'm getting a strange v-shaped volcano plot from limma. The data matrix is small - 400 observations, for two conditions - cond1, cond2,cond2 (so only have replicate for cond2).

The plot has a distinct v shape:

http://imgur.com/a/2NDXw

Has anyone seen something similar?

Many thanks.

It's hard to say because you don't put the code you used for the analysis. The most likely scenario is that limma's estimate of the prior degrees of freedom is infinity (see fit$df.prior). If you didn't set trend=TRUE when running eBayes, this will mean that the t-statistic is directly proportional to the log-fold change between groups (as variances for all genes will be the same). This causes one of the lines on the volcano plot. I'm not sure what the other line is caused by - possibly a different variance estimate with robust=TRUE.

P.S. Split your tags so that limma is a separate tag, otherwise the maintainers don't get notified.