RNA-Seq analysis with DESeq2: showing two conditions do *not* significantly differ
2
0
Entering edit mode
Ndimensional ▴ 20
@vlaufer-14169
Last seen 16 months ago
United States

Hello bioconductor,

I am working with a collaborator who is interested in showing that two states that people in the field might expect to differ in fact do not differ. I am writing to ask what are the most convincing ways to do this both visually and statistically.

Here are my thoughts:

Visually, I think if we took the gene signature that would be expected to differ most strongly, then we showed that those genes in fact do not differ using a heatmap and a distance matrix, that could be a good way to make the argument visually in a context that is relevant in the field.

Statistically, we might think about using permutation-based testing to show that the p-value distribution of the conditions does not significantly differ from random permutations of the labels. This option seems more rigorous numerically but I think it is also less likely to be immediately grasped by a larger audience.

What other approaches could be fruitful to take?

Thank you!

deseq2 RNA-Seq uniform distribution data visualization • 908 views
ADD COMMENT
3
Entering edit mode
@mikelove
Last seen 1 day ago
United States

If you read over the DESeq2 2014 paper you'll see we created a method for this, and it's also discussed in the vignette. Search for altHypothesis.

ADD COMMENT
0
Entering edit mode

my man - that webpage is practically my backyard. Don't know how I missed this feature before. Thank you. I'll dive in there and into the documentation itself. Appreciate your great work. VL

ADD REPLY
1
Entering edit mode
@steve-lianoglou-2771
Last seen 21 months ago
United States

You could also set altHypothesis = "lessAbs" in your DESeq2::results() call to identify genes whose differential expression is below some threshold.

Check out the Tests of log2 fold change above or below a threshold in the DESeq2 vignette.

Edit: woops, looks like Michael beat me to the punch

ADD COMMENT

Login before adding your answer.

Traffic: 488 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