Entering edit mode
manuelw
•
0
@manuelw-12775
Last seen 7.1 years ago
I would like to add a pseudocount of 0.0001 to genes with 0 counts. I can't add 0.0001 to the count table, because DESeq2 requires integers as input. Is it possible to replace the zeros with pseudocounts in the DeseqDataSet and then do the analysis?
Greetings
Manu
I want to perform differential expression analysis for each genome position (so "gene" was a simplification) of a RNA modifying enzyme. In the treated condition a position with n>0 counts means that this is a possible enzyme interaction site and it is checked for differential expression against the untreated condition.
I was wondering if pseudocounts would lead to more differentially expressed positions and want to test for that.
I don't know exactly what's going on, but pseudocounts aren't an option with DESeq2. Maybe check out the paper to get an idea what we're doing:
http://doi.org/10.1186/s13059-014-0550-8
I know what you're doing :-)
It's just that I found a paper doing similar stuff to that what I want to do. They wrote, that they'd used pseudocounts with DESeq2. I contacted them, but I'm a little impatient and was also wondering if there is a general way to use pseudocounts in DESeq2. That's why I posted here.
The term "pseudocount" is usually used in the context of adding a small constant to a number which might be zero before taking the logarithm. Hence, pseudocounts are irrelevant unless you are taking the logarithm of the counts, which is not the case in DESeq2. The paper you're looking at might have used the term "pseudocounts" to refer to estimated counts from RSEM or similar tools, even though that is not a generally accepted usage of the term.