My sample is RNA-Seq data from human Esophageal tissue, 9 control groups（normal）, 2 experimental groups（Esophageal cancer），The result is Figure 1.
I use featurecounts to count genes, this is my counts array
Those lines are from genes with very low counts of genes, e.g. 2 counts in the treated group, 1 count in the control group. If you want to remove them you can perform some minimal filtering, e.g.:
keep <- rowSums(counts(dds) >= 10) >= x
dds <- dds[keep,]
Where you choose x to be a small number of samples.
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
Will removing these low counts have any bad effect, will I ignore the low-expression genes?
It's pretty hard to achieve power for DE when there are say, only 2 samples with counts above 10. So setting
x <- 3will be fairly conservative in terms of FNR.Ok, thank you very much！