Detection of (novel) lnc-RNAs in 'normal' RNA-Seq data

Question

[aggregatibacter]

Hi folks,

I have several human RNA-Seq datasets (libraries prepared from total RNA using the TruSeq Stranded Total RNA kit from Illumina) that I would like to search for lnc-RNAs (both known and unknown).

• Is this possible? I read that some library prep kits do not keep all lnc in - but this kit does not do poly-A enrichment, so it should be ok?

• Known lnc should not be a problem with normal alignment procedures. How would you suggest to hunt for novel lnc-RNAs?

Cheers!

Answer 1

[igor]

It's a total RNA kit, so it should be fine for lncRNAs. GENCODE has many lncRNAs annotated. You also should look into MiTranscriptome (large-scale ab initio transcriptome meta-assembly from 6,503 RNA-Seq libraries): http://mitranscriptome.org/

Hunting for novel lncRNAs is not trivial. They are not highly expressed, so it's hard to distinguish the signal from the noise. If you identify a potential novel transcript, you need to determine if it's a lncRNA using metrics like size, coding potential, and nearby genes. lncRNA blog keeps a reasonable log of available tools:

• http://www.lncrnablog.com/category/technology/methods/lncrna-discovery/
• http://www.lncrnablog.com/category/technology/methods/analysis-pipelines/

[aggregatibacter]

Thanks for pointing me to the mitranscriptome, I had not seen that project before...

[igor]

It seems really promising. Not sure why it's not more widely known.

Answer 2

[i.sudbery]

Checkout this review: Predicting long non-coding RNAs using RNA sequencing.

In a nutshell, you are going to want to apply a transcript assembly algorithm like cufflinks or stringtie (I just tried out stringtie, and it ran in 40 minutes where as cufflinks took 50 hours, so I recommend that), and then filter like crazy.

[aggregatibacter]

Many thanks for the quick and helpful advice..

[Friederike]

What kinds of "crazy" filters do you have in mind?

PS: The link is broken.

[igor]

Some options: length, expression level, coding potential