Closed chubukov closed 8 months ago
Hi @chubukov, thanks for pointing this out.
I just tested the example you provide in #237 in the original R DESeq2 package. The resulting behavior is to replace outliers even if it leads to the gene only having zero counts, in which case the baseMean
is then set to 0 and all columns in the results dataframe to NaN
.
here and here there is a filter for not doing outlier replacement or refitting if the result of the replacement would be all zero counts.
Why is this the right thing to do? If I have zero counts in all but one sample, and a large count in the single outlier sample, why wouldn't the right result be to remove the outlier?
Is this consistent with what the R DESeq2 package does?