DESeq2 pvalues overly conservative
1
0
Entering edit mode
gdolsten • 0
@bf72196f
Last seen 1 day ago
United States

Hi, I am running a DESeq2 analysis for some data. I have two different conditions in my data which should have identical RNA-seq at day 0 (I am doing a time course experiment). However, when I run them through DESeq2, I get a hill-shaped pvalue distribution (plotting non-adjusted p-values), shown below. Do you have any idea why this might be?

DESeq2 • 90 views
1
Entering edit mode
@james-w-macdonald-5106
Last seen 4 hours ago
United States

In my experience, that is relatively easily fixed by using svaseq from the sva package to generate some surrogate variables to account for unobserved variability.

0
Entering edit mode

Thanks! Great suggestion. I understand the idea of generating orthogonal vectors for your observations if you have a binary comparison. Do you have any intuition for how many surrogate variables to include, or to understand if this is somehow overfitting/how much variance is being explained?

Traffic: 337 users visited in the last hour
FAQ
API
Stats

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