Differential expression of microRNA with no replicates

Question

[Saad Khan]

Hi,

Various differential expression analysis packages for RNA-seq (like edgeR and DEseq) provide support for analysis when there is no replicate available. I was wondering if I can use edgeR or DEseq to do the same for microRNA ? And if so do I need to do any p-value correction further to get a more valid result.

[Devon Ryan]

Note that while these tools allow you to use unreplicated data, the results are questionable. This isn't the fault of the tools, but rather is innate to having unreplicated data. You're also likely to get somewhat more reliable results with GFOLD in these cases.

Answer 1

[Sean Davis]

These packages will work with a matrix of counts. They need not be mRNA. Multiple-testing correction is always needed when multiple tests are done, but I don't think you need to deviate from the standard workflow, otherwise.

[Saad Khan]

So once I have results from Gfold what should I do or how should I go to refine my results.

[Devon Ryan]

I assume you meant this in reply to me. Anyway, you'd want to filter by log2fdc. That's as good as you can do with unreplicated data.

[Saad Khan]

That's what I want to know how to choose the optimum log2fdc values

[Devon Ryan]

There is no such optimal value.

[Saad Khan]

So on what basis do you set a filtering criteria for log2fdc values

[Devon Ryan]

Look at the data and see if any particular value makes sense. There is no objective criterion for this.