A couple of weeks or months ago, a question was posted regarding DESeq2's behavior when dealing with samples with one extreme outlier. In that case, DESeq2 identified that gene as DE, although there was only 1 out of N samples with extreme counts.
Michael Love suggested a way to force DESeq2 to look at all replicates.
I've been searching for the post for a while using both Google and the forum search but I can't find it.
Can somebody post the link, or maybe repeat the explanation here?