vst with output of edgeR::calcNormFactors
2
1
Entering edit mode
ATpoint ★ 4.0k
@atpoint-13662
Last seen 3 hours ago
Germany

Hi Mike,

hope you have a minute to clarify:

I would like to plug in existing size factors into vst like

sizeFactors(dds) <- edgeR::calcNormFactors(count.matrix) and then vst(dds)

in order to use the exact same factors I used for DE testing with edgeR.

While technically possible I'd like to know whether the direct output of calcNormFactors is appropriate here. Thanks for your time!

DESeq2 • 1.1k views
ADD COMMENT
1
Entering edit mode
@mikelove
Last seen 10 hours ago
United States

Size factors in DESeq2 should be on the scale of the counts and account for multiplicative differences in the counts across samples: if size factor for sample j is 2x bigger than for sample i, then counts can be expected to be 2x bigger for j compared to i across null genes.

To my understanding, I think you have to multiply the output of edgeR::calcNormFactors by the column sums of the counts matrix to get something like what you get from DESeq2::estimateSizeFactors.

ADD COMMENT
0
Entering edit mode

Thanks, makes sense. I get a bit into trouble with the less than 'nsub' rows with mean normalized count > 5, since the magnitude of the factors from edgeR are completely different than the ones from DESeq2 (much larger with edgeR when multiplying with colSums, maybe try to scale by 1e+06), but this is not DESeq2-specific anymore, therefore won't bother you with it. Will try to figure out sth. that works for me, thanks for your answer!

ADD REPLY
0
Entering edit mode
@gordon-smyth
Last seen 13 minutes ago
WEHI, Melbourne, Australia

If you have an DGEList object then you can use effectiveLibSizes(dge) to get library sizes taking into account norm factors from TMM or otherwise. (Renamed to getNormLibSizes(dge) in edgeR 3.41.1.)

ADD COMMENT

Login before adding your answer.

Traffic: 737 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6