What is cut-off for significance in DE analysis in sc-RNA seq data

Question

[shome]

Hi, Generally |LogFC| > 2 and p-value < 0.05 is considered for determining the cut-off for significance while carrying out differential expression analysis in bulk RNA-seq data analysis.I am wondering what is the cut-off values considered in case when carrying out DE analysis in single-cell RNA seq data analysis ?

[rpolicastro]

Just a note, the common p-value cutoff is just p-value < 0.05. If your cutoff was -log10(p-value) < 0.05 you would actually be selecting for p-values great than about 0.3.

[Istvan Albert]

also to add to this, a LogFC > 2 means a four-fold change, I would say that is an overly ambitious effect size.

in general there is no such thing as "what value should be considered" -

instead, it works like this, when people have plenty of DE genes they set generous cutoffs boasting of how "strict they are",

but then when nothing comes up DE they are scraping by pushing those filtering conditions as much as possible, it is just a reality of what happens in the field. Some data is a lot more precious and difficult to get than others. And when you have you got to make do with what you have.

[ATpoint]

Adding on this, due to the lo counts per gene and cell in single-cell data the dynamic range of the logFCs is generally lower than in bulk data while pvalues (at least on single-cell level) are inflated.

[shome]

Thank you for pointing it out !