Intronic reads, ongoing transcription and co-transcriptional splicing

[adameur]

Hi all

We have studied intronic reads in total RNA-seq data and have some new and interesting results. Briefly, we found that nascent (ongoing) transcription and co-transcriptional splicing can be detected by RNA-seq. For more details see:

http://dx.doi.org/10.1038/nsmb.2143

This could explain the large proportion of intronic reads seen in many RNA-seq datasets..

Hope you like!

Adam Ameur

[steven]

That looks very exciting indeed! I was looking forward to such a study..

[adameur]

Being an author of this paper I’m obviously biased, but I’d like to highlight some aspects of this of this work that I think is of general interest for the design and analysis of RNA-seq data:

1) A large fraction of reads in total RNA-seq data are intronic, up to 40% in some samples. These intronic reads can be used to study the rate of nascent (ongoing) transcription, while exonic reads mainly measure mature mRNA levels. So intronic reads are not garbage, and why don’t try to use them to get a better view of the transcriptional activity in the sample...

2) Analysis tools should take into account the high proportion intronic reads detected in RNA-seq data sets. Also, it may be possible to develop new bioinformatics tools to study levels of ongoing transcription and splicing (based of the intronic read coverage).

3) PolyA+ RNA-seq gives very different results compared to total RNA-seq. For polyA+, a much higher fraction of reads are mapped to exons. So if you are mainly interested in mature mRNAs and don’t want to waste reads on the introns, it might be a good idea to go for polyA+.

These were just some points I’ve been thinking about. It would be great to hear about your experience with intronic reads and what you think about these ideas.

[pmiguel]

I remember an old (pre-next gen) cDNA library construction paper that recommended doing cytoplasm preps (that is, removing the nuclei from cell lysates) prior to isolating total RNA to avoid pre-spliced mRNAs.

--
Phillip

[JakobHedegaard]

Analysis of totalRNA-Seq data

Hi Adam Ameur (and other readers of this thread),

Found this thread after reading your article (Ameur et al 2011).
I am presently challenged with the analysis of RNA-Seq data from rRNA depleted total-RNA from paired fresh frozen and FFPE human tissues. As observed in other studies, I also finds a large fraction (40-60 %) of the reads mapping to intronic regions (80-90% for FFPE).
To generate expression profiles of the mature transcripts (based on the exonic reads) I have been considering to correct the total exonic reads for reads mapping to pre-mature exonic regions (defined from the intronic reads). The reads mapping to exon-exon junctions could perhaps be used as an indicator of the true mature expression level?

Appreciate commnets.

Cheers,
Jakob

[adameur]

Hi Jakob,

Yes, I think it makes sense to use the exon-exon reads to quantify mature mRNAs. That is if you have sufficient number of exon-exon reads for the statistical analyses.. Interesting that you get more intronic reads in the FFPE samples, any ideas why?

Also I can mention that we have been working on a cytoplasmic RNA prep to enrich for mature mRNAs. We have seen that cyto-RNA gives much lower levels of intronic reads compared to both total and polyA+ RNA, so maybe that could be something for future experiments..

Cheers

Adam

[shi]

Hi Adam Ameur (and other readers of this thread),

Found this thread after reading your article (Ameur et al 2011).
I am presently challenged with the analysis of RNA-Seq data from rRNA depleted total-RNA from paired fresh frozen and FFPE human tissues. As observed in other studies, I also finds a large fraction (40-60 %) of the reads mapping to intronic regions (80-90% for FFPE).
To generate expression profiles of the mature transcripts (based on the exonic reads) I have been considering to correct the total exonic reads for reads mapping to pre-mature exonic regions (defined from the intronic reads). The reads mapping to exon-exon junctions could perhaps be used as an indicator of the true mature expression level?

Appreciate commnets.

Cheers,
Jakob

Your intronic reads were likely coming from those unspliced transcripts because you used total RNA for sequencing in your experiment. However, many of these intronic reads may not be useful because the corresponding intronic regions were spliced out and were not included in the mature RNA.

[lyndaben]

We are seeing a big difference according to the method of cell lysate storage and RNA isolation: With Qiazol and miRNeasy-performing RNA-Seq we get lots of intronic reads that looks very much like nascent transcripts (for a subset of genes) (See Bhatt et al., 2012, Cell 150, 279-290); Using RLT, gDNA eliminator and RNeasy-almost no intronic reads in the same genes.

It is most apparent in large genes with multiple exons and alternative splice options

Lynda