miRNA predictions from large set of long non coding RNA

Question

[Bioinfonext]

I have de novo assembled transcript using Trinity. After annotation with nr database and Coding potential filtering, I have around 10,000 transcripts which can be considered as non-coding transcripts.

From these transcripts, I want to predict how many of these form hairpin loop structure and what miRNA can be generated from these transcripts. I will be thankful for your suggestion.

some people suggest using, RNAfold from the ViennaRNA package and mfold for hairpin loop structure prediction, but I am not sure How these work for large set of transcripts? and further after hairpin loop structure prediction how I can predict the miRNA sequences from all these transcripts.

Thanks Yogesh

[cpad0112]

try mirdeep2 workflow Bioinfonext

[lieven.sterck]

do you have any info on how the RNA was extracted that was used as input for your assembly?

When looking for small RNAs people usually use a 'special' protocol to also capture the small(er) RNAs

[Bioinfonext]

Thanks, this is not small RNA sequencing. I have extracted total RNA and mRNA was used for sequencing using poly A tail selection.

[i.sudbery]

A second vote for mirdeep2.

If your aim is to predict primary miRNA precursors, you should be aware that these precursors are genearlly very unstable - they are rapidly degraded after the mature miRNA has been processed out of them .This is particularly true in polyA seq, where once the miRNA is processed out of the transcript, the polyA tail will now be detached from the rest of the transcript due to the cleavage steps in miRNA processing.

[Bioinfonext]

Thanks for your response. But still people detecting long non-coding RNA from mRNA sequencing and these long non coding RNA is precursor to miRNA. They may not able to capture all the precursors but still lots of long non-coding is reported from mRNA sequencing.

[lieven.sterck]

as far as I know long-non-coding RNAs are a totally different thing then miRNA precursors. Yes you absolutley can detect those lncRNAs from the given data but just as i.sudbery I seriously doubt the success rate for finding miRNA precursors.

[Devon Ryan]

these long non coding RNA is precursor to miRNA

Only if you're talking about the whole cluster. Sure, those transcripts are long enough, but they don't stick around terribly long. These are also not typically considered lncRNA.

[i.sudbery]

Where miRNA precursors are annotated in the genome builds, I think they generally do have the GENOCDE lncRNA biotype, but the fact that few of them are annotated tells you something about the difficulty in finding them.