Question

DESeq2 One-way ANOVA with 3 sample groups?

1

Entering edit mode

alex.pelletier ▴ 10

@alexpelletier-6758

Last seen 9.1 years ago

United States

I'm trying to use DESeq2 to perform something like a one-way ANOVA on 3
groups of samples. I've seen references to DESeq2's capability of
performing similar ANOVAs such as this forum thread:
http://seqanswers.com/forums/showthread.php?t=16539

However, I can't seem to find anything in DESeq2's manual about how to
do this. I have my data inputted successfully, and I've been able to
perform pairwise comparisons using DESeq2 but I haven't been able to
extract one p-value across 3 different groups. Is there somewhere that
explains this functionality? Or some sort of example code run through?
Thanks for your help.

deseq2 Anova • 11k views

ADD COMMENT • link updated 3.2 years ago by Michael Love 43k • written 10.3 years ago by alex.pelletier ▴ 10

score 3 · Answer 1 · 2014-09-19

3

Entering edit mode

Michael Love 43k

@mikelove

Last seen 11 hours ago

United States

The likelihood ratio test will give you a single p-value for the removal of a variable from the design.

For example, if 'group' is a factor in colData(dds) with three levels:

design(dds) = ~ group
dds = DESeq(dds, test = "LRT", reduced = ~ 1)
results(dds)

See the likelihood ratio section of ?DESeq and ?nbinomLRT for more information.

ADD COMMENT • link 10.3 years ago Michael Love 43k

0

Entering edit mode

@Michael Love,

I have a follow-up question here on DESeq2 One-way ANOVA with 3 sample groups. I honestly speaking did not understand well how this One-way ANOVA works here. Basically, my question is I would like to do get differentially expressed genes in one condition compared to the other two conditions. This means I have three groups(Cancer_Staging_1, Cancer_Staging_2, and Cancer_Staging_3). I want to see genes specifically expressed in Cancer_Staging_1 considering Cancer_Staging_2 and Cancer_Staging_3as a comparison group at a time.

Best, AD

ADD REPLY • link 3.2 years ago adR ▴ 40

0

Entering edit mode

You can do two pairwise comparisons and take the intersection.

ADD REPLY • link 3.2 years ago Michael Love 43k