Indpendent Filtering in DESeq2

Question

[EVR]

Hi, I have a RNAseq data and I generally perform the DESeq2 analysis setting "Independent filtering=T". Is it sensible to filter the out transcripts/genes which has raw counts < 20 and then apply "Independent filtering=T"? Kindly guide me.

Thanks in advance

Answer 1

[Devon Ryan]

There's not much of a point in prefiltering if you're going to independently filter anyway.

[Carlo Yague]

The only advantage I see is to end up with a smaller table and save a bit of memory from the start. This however would probably not be noticeable for most analysis on a decent computer.

[Devon Ryan]

Indeed, if someone had enough samples to worry about memory then I expect the overall performance of DESeq2 would be abysmal (this is a case for limma, then).