Question

DeSeq2 - design formula for enrichment on combinated conditions AB, not A, not B

0

Entering edit mode

johannesdolberg • 0

@johannesdolberg-24282

Last seen 2.3 years ago

United States

Hi,

A very simple question (I hope) from a Deseq2 newbie.

Considering cofactors A and B.

If you do:

Experiments containing cofactors A and B.
Experiments containing cofactors A only.
Experiments containing cofactors B only.
Experiments containing no cofactors.

Is it possible to write the design formula - or tweak DeSeq2 in some way - so one can extract genes enriched in the experiments with both cofactor A and B compared to the experiments with cofactor A only and cofactor B only?

Best regards, Johannes

conditions design DESeq2 Synergy • 849 views

ADD COMMENT • link updated 2.4 years ago by Michael Love 41k • written 2.4 years ago by johannesdolberg • 0

0

Entering edit mode

Well, seems what I needed to look for was the term "synergy". Believe I've found an answer here.

ADD REPLY • link 2.4 years ago johannesdolberg • 0

score 0 · Answer 1 · 2021-12-20

We don't have the ability to find these with a design formula + results() combination. You can find cases where there is more than additive effects (on the log scale) with an interaction term.

When users want something to be significant in one comparison but _not_ significant in another, I recommend using lfcThreshold=x and altHypothesis="lessAbs" for a threshold x and then just looking at the intersection of the FDR-bounded sets of interest.