I have run my initial analysis without filtering for low-expressing genes, given DESeq2's independent filtering procedure. Due to the low number of significant genes returned when using an FDR, we have opted to use an unadjusted p-value cutoff so that we can generate a gene list of sufficient length for GO analysis. Because DESeq2's independent filtering informs the FDR adjustment, is it recommended that I remove low-expressing genes prior to running DESeq2 if I'm using an unadjusted p-value? Or is the independent filtering threshold considered when using an unadjusted p-value cutoff as well? I am using DESeq2 v1.32.0.
Thanks for your help, Emily