Question: DESeq2 interaction padj NA values
gravatar for meshi.barsheshet
5 months ago by
meshi.barsheshet0 wrote:

I carried out RNA-seq for WT strain and a mutant strain of gene X, grown under two conditions (no treatment, treatment). To extract the effect of gene X on changes in gene expression between treatment and no treatment, I applied DEseq2 with the "interaction" option.

My design : ~genotype + condition+ genotype:condition

For quite a few genes I got substantial log2FC values for the interaction term but the padj values are NA.

For the same genes the individual comparisons yielded statistically significant results [e.g. WT(treatment vs. no treatment) and treatment( WT vs. mutant of gene X).

Here are two examples, one with statistically significant padj for the interaction, and another one with NA. Numbers are for log2FC and padj

Gene    Basemean    Treatment (WT vs mutant)        WT (treatment vs no treatment)      No treatment (WT vs mutant)  Mutant (treatment vs no treatment)Interaction
Gene A  2201    -3.96, 0.000    -4.13, 0.000    -0.57, 0.769    -0.743, 0.145   -3.387, 0.000
Gene B  65.523  -3.310, 0.001   -3.250, 0.002   0.101, 0.993    0.161, 0.911    -3.411, NA

I believe I can conclude that gene X has an effect on the expression change of gene A between treatment and no treatment. What would be the conclusion for gene B, which has a substantial log2FC for the interaction but the padj is NA? Of note, it is clear that without an active gene X there is no change in gene B expression between treatment and no treatment (log2FC of 0.161).


ADD COMMENTlink modified 5 months ago by swbarnes29.4k • written 5 months ago by meshi.barsheshet0

Please use the formatting bar (especially the code option) to present your post better. You can use backticks for inline code (`text` becomes text), or select a chunk of text and use the highlighted button to format it as a code block. I've done it for you this time.

ADD REPLYlink written 5 months ago by _r_am32k
gravatar for swbarnes2
5 months ago by
United States
swbarnes29.4k wrote:

It might be that some of the p values for some of the comparisons are too poor for the software to conclude they are difference with a good p-value.

ADD COMMENTlink modified 5 months ago • written 5 months ago by swbarnes29.4k
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1795 users visited in the last hour