high throughput sequencing of microRNA

Question

[bernaines]

Which is the appropriate coverage for high throughput sequencing of microRNA? 1.000.000 reads per sample???

[Devon Ryan]

Depends on your goals and how much piRNA/rRNA/tRNA you have in the cells and how you prepare the libraries.

[Chirag Nepal]

But still, 1,000,000 reads per sample is low, in general. I think 10M reads is reasonable. Most importantly, library prep is important as also mentioned by Devon.

[bernaines]

I used the "truseq small rna sample prep kit (Illumina)" for library prep, and now I'm confused about which kit use for the Sequencing, I was thinking of using the MiSeq Reagent Kit v3, which generates 25M OF REads, and I have 24 samples, so.. I will be having 1 M reads per simple..and I don't know if that is sufficient

Answer 1

[Devon Ryan]

You might have a look at this paper, which has a couple nice graphs that will help you guide how much depth you need.

So you might try to go for a few more reads if possible, but a million might not be terrible, particularly if you have 24 replicates. Again, this will depend entirely on the tissue of origin. In sperm, for example, there's relatively little miRNA but a lot of piRNAs, so if you were interested in miRNAs there you'd want to sequence more. This will also depend a bit on the kit you use and how degraded your samples are. If the rRNAs start breaking down then you're going to have a bad time trying to get much out of such low depth (but if they haven't then that might not be too bad a target depth).

[bernaines]

Thanks for the help! and for the article!

I am working with muscle tissue in cattle, my goals are to identify the microRNAs already annotated for this tissue and to discover novels microRNAs.

I tried to check in the mirbase the number of microRNAs annotated for muscle tissue in cattle, but there was not information about that in this data base.

My samples were not degraded, because I checked the integrity in the Bioanalyzer for the 24 samples, and all of them were with a good RIN (7 -8).

I have 12 replicates per treatment (2 treatment) and I was thinking in put the 2 pools in the same flow cell in the Miseq, using the Miseq Kit v3, however the coverage will be less than 1M reads per sample.

Do you think that this is right? Do you generally make the sequencing in Hiseq or Miseq? And wich Kit do you use?

[Devon Ryan]

We only have access to a Hiseq, so we just throw everything on one lane on it (yes, this is overkill). I haven't a clue which kit is used since I have others make the libraries for me :) Hopefully Chirag Nepal can provide some advise there. For actually discovering new miRNAs, the power is actually derived from the sum of mapped reads across all of your samples, so I actually wouldn't be too worried there. The only problems you might run into would be (A) rRNA or some other species you don't want sucking up too many reads or (B) lack of power on low expressed isoforms due to low per-sample read counts. Of course the only way to know if those will be a problem is by doing the experiment.

[Chirag Nepal]

Sadly, don't have much idea on library preparation. I also have other people do RNA library prep for us.

Answer 2

[Chirag Nepal]

It all depends upon what are you looking for? With library as low as 1M,

1. It is difficult to find novel mirnas, as novel mirnas are generally found when sequencing depth is increased. Simply because other studies might have already reported other mirnas that are highly expressed in that samples/tissues/celllines
2. Detecting Iso-miRNAs or moRNAs even for annotated mirnas will be tough