Making small RNA lib. - do I need DNAse?

[JonB]

Hi,

I have isolated RNA smaller than 200 nts and I am going to sequence on the Illumina platform. I did not use DNase during the isolation step.
My question is if this is really necessary?

I have run the samples on a gel and there are no traces of genomic DNA, but of the gels are not that sensitive. Also I worry that DNase treatment might just chop up the DNA into smaller fragments and that will actually increase the contamination in my library! Further, the library prep (ScriptMiner from Epicentre) favours RNA over DNA (I think). Finally I am afraid that I will loose much RNA when I clean up after DNase treatment.
I also think that any DNA contamination is possible to identify bioinformatically (?).

Therefore I feel that using DNase is only to please reviewers and not really necessary.

Any thoughts, comments, or links to studies testing this would be greatly appreciated.

Jon

[JamieHeather]

Some bits of gDNA sequence you would be able to ID, but not all - a read that covers part of an exon could easily be derived from either DNA or RNA, with no features to tell otherwise.

If you're really opposed to it you could just measure the amount of DNA you have in your RNA samples, on a qubit or similar. Or you could do a PCR for an amplicon that either would only exist in gDNA, or would have different sized bands once transcribed and processed.

From my own experience (with different kits, for different purposes) I always get some gDNA contamination of my RNA preps. I don't lose much from my DNase treatment, although personally I'd rather have less total RNA and know it's not contaminated than increase my yield and possibly distort my results.

[JonB]

Thanks!

I have almost decided that I will skip the DNase. I think that any expressed RNAs will stand out as non-random compared to any DNA contamination. Also I have several samples from the same tissue types so I can compare expression between samples.

But thanks for the advices.

[MasterHammerMan]

Dear JonB,

I have two questions:

1. Did you see expression differences for samples with DNase digestion and without, especially for small RNA-seq?

2. What are your experiences using the ScriptMiner from Epicentre?

We are planning to do a small RNA-Seq run on a MiSeq system with ScriptMiner™ Small RNA-Seq library preparation (Epicentre) and we ask us the same question like you, do we really need a DNase digestion!?

Thanks and best regards,
Paul

[JonB]

Hi Paul,

We did not sequence samples with and withouth DNase treatment, so I am not sure how the expression profiles would look.

We didn't use DNase treatment on our samples, but we had a company do the library prep (Vertis Biotech) and I am not sure whether they used ScriptMiner. (I have later failed to make smallRNA libs with that kit because I got only adaptor dimers. I think it can be a challenge to find the right amount of adaptors - but I have only tried once though…).

But we did get a lot of short reads that mapped in both directions to the genome and which we could not determine if they were true transcripts or from DNA contamination. I think that having one additional sample with DNase treatment would have helped us a lot in this respect. Otherwise, if you can sequence multiple samples without DNase and then only choose transcripts which are consistent in all samples, I would say they are not contamination.

From my experience with smallRNA sequencing of non-model organisms, it is a challenge. There are no standard and well-tested softwares to use, and you find a lot of short reads that you cannot say if they represent true transcripts, or contamination, or just some random expression, or whatever. You can detect already known smallRNAs, but novel smallRNAs are hard to identify. At least without additional experiments like knockout or something.

I apologize for this negative answer.

Good luck!