Hi,

When subsetting an EList object as output by voomWithQualityWeights() there is a list element $sample.weights. When subsetting using the single bracket method, this element returns the initial vector, instead of one that matches the subset. This doesn't seem to be the correct behaviour. Is this able to be fixed?

y <- matrix(rnbinom(10000,mu=5,size=2),ncol=4)
d <- DGEList(counts=y, group=rep(1:2,each=2))
v <- voomWithQualityWeights(d)
v1 <- v[,1:2]
length(v$sample.weights) == ncol(v)
length(v1$sample.weights) == ncol(v1)

Cheers,

Steve

This does seem like a bug, though subsetting the output of voomWithQualityWeights is unusual. Is there a reason why you use samples 3 and 4 in voomWithQualityWeights but not for the rest of the analysis?

Yes, that's a fair point. The problem is that the list component sample.weights returned by voomWithQualityWeights() is not a recognized component of an EList object, and hence it isn't treated properly by the subsetting operations. The recognized components can be seen from help("EList-class"). EList objects assume that all vectors corresponding to samples (columns of the object) are included in the targets data.frame. 

I have now changed voomWithQualityWeights() so that sample.weights are returned as a column of the targets data.frame. This ensures it will subset properly. If you want to extract the sample weights in future, you will need to use $targets$sample.weights.

The revision has been committed to limma 3.34.5 (release) and limma 3.35.5 (devel).