Hello, there,
Our cells were treated in two tough conditions (A and B), and as a result, numbers of the survival cells were pretty low, particularly for condition A. The total mapped reads for condition A was about 10% of that for condition B. I wonder if I can still run DESeq2 for differential analysis between A and B.
Thanks a lot!
C.
What are the number of reads for the two groups (in millions)?
I think this is a problem for any kind of DE analysis, because you can get a count of 20 in condition B, but a 0 in condition A under the null. We discuss this as a pathological case in the DESeq2 paper. One way to deal with this would be to filter out genes unless you see a positive count for all samples. This isn't ideal but otherwise you are subject to the problem that 0's which are compatible with the null will look like significant differences.
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
How about variance-mean dependence for low-reads sample? Does it still hold?
Yes VST stills works if that is what you are asking?