miRNA Relevant Count Expression and Normalization

[HughMyron]

Hello,

I am new to this so please bear with me if my question is too vague.

I have some small RNA-Seq (miRNA) sequencing data that I have aligned and mapped and have raw counts for, and I have a two-part question:

1) Does anyone have any literature that describes the "minimum relevant" count that a certain miRNA would need to be expressed at to possess its silencing activity? Obviously, this is relevant to the system/experiment, so any general literature on the subject would be great. (Google was not my friend today)

2) What is the best normalization technique for a small sample size (n=150) of miRNA small RNA-Seq experiment?

Thank you in advance.

[Erlkoenig]

1) I wouldn't say there are any hard and fast rules about the minimum number of reads for determining biological significance. Your ability to identify low-abundance transcripts depends very much on the sequencing depth obtained in your experiment. This applies for all RNA-seq, whether you're looking for miRNA, mRNA, etc. A "rare" read could actually be highly expressed in only a small subset of your sample, but nonetheless very important. Try this paper and associated references:
Tarazona, S., García-Alcalde, F., Dopazo, J., Ferrer, A., & Conesa, A. (2011). Differential expression in RNA-seq: a matter of depth. Genome Research, 21, 2213–2223.

Of course, the more reads you have of a particular transcript, the more convincing your ability to determine differential expression.

2) If library size is relatively close between samples, reads per million is a good place to start for miRNA. Don't get carried away with normalization unless you have a specific reason for doing so.