pri-miRNA sequencing

[NicoBxl]

Hi,

Is it possible to detect pri-miRNA from RNA-seq data ? Is there any studies on this subject?

Thanks,

N.

[NicoBxl]

nobody knows ?

[SOLiD_noobie]

If you search seqanswers, I think you will find one paper and a database of about 80 pri-miRNAs http://www.biomedcentral.com/1471-2164/9/564 . You could use that database with your RNA-seq data, but I do not personally have a way to identify all pri-miRNA data in an RNA-seq experiment.

What are you trying to do?

Brian

[Gators]

Since primary miRNAs are nothing more than Pol II transcribed RNA transcripts, they should show up in RNA-seq datasets. If you have size-selected for only transcripts greater than ~90nt you should be able to use mirBase's hairpin fasta file to align your reads and pull out primary miRNA reads. Unfortunately this will only give you pri-miRNA reads that align to the hairpin, and the primary miRNA consists of much more than the hairpin, but it would be a start. If you did not size select, you will probably also pull out pre-miRNA reads and possibly even mature miRNAs if you have reads that are as small as ~20nt.

[SOLiD_noobie]

If you size select greater than 90nt you will get pre and pri miRNA reads mapping to the hairpin. In addition to that, you will still get some short mature reads that escape the size selection (this is my experience using ABI SOLiD data). If you set a minimum cutoff for the read match, that can eliminate the mature data, but you are still stuck trying to differentiate between pre and pri with the hairpin data.

We have looked at the sequence immediately adjacent to the hairpin sequence (100-300nt) and exclude anything defined as an exon, to see if this gives a clue about pri-miRNA expression. I have not seen any results that convince me we are seeing much pri-miRNA signal in our WT libraries.

[Gators]

That's interesting, I am a bit surprised you are seeing pre miRNAs in your sequencing data when you size select greater than 90nt. I could've sworn that pre-miRNAs are closer to 75nt in length or even smaller.

Ah well. Bottom line is that there is no easy way to do it.

[SOLiD_noobie]

Human miRNAs average about 85 nt. Size selection if far from a digital technology. We have run WT and small RNA experiments with the same sample. When we check strong mir signals that show up in our WT data, it almost always looks like mature sequence (short ~25nt reads that map to mature and mature* position on the hairpin).

I agree with your bottom line, but I think with a little work it could be done. I suspect part of the issue will be the half-life of pri-mirna compared to other signals in a sample.

[NicoBxl]

My purpose is to detect a pri-miRNA which "coded" for 5 pre-miRNA ( there are very close from each other ) so my idea was to align the reads against the part of the genome where this cluster is coded to find the start and end of the pri.

maybe inhibit Drosha and then RNA-seq ?