Ribo-depletion

[Thomas Sandmann]

Hi, 

I am dealing with RNA-seq data from libraries that were generated by depleting rRNAs (rather than e.g. by enriching for polyA rRNA).

I noticed that the efficiency of the ribo-depletion varies, as judged by the fraction of reads mapping to rRNA loci. 

How do varying rRNA contaminations affect Salmon's quantification results? E.g. should I filter the reads and remove those that map to rRNA transcripts before mapping and quantitation with Salmon?

Thanks a lot for any feedback,

Thomas

[smount]

I think that the best strategy is to include rRNA in your transcriptome. That way, reads that map to rRNA can be assigned to rRNA (rather than being assigned erroneously to some mRNAs that might have a chance similarity).

[Thomas Sandmann]

Thanks a lot for your quick response! Yes, can definitely ensure that at least one representative of the rRNA transcripts is in the reference.

Given that the abundance of rRNA reads is a function of the depletion efficiency, I guess the TPM estimates should be interpreted with caution.

For differential expression analysis, I am planning to renormalize by calculating size factors after removing the rRNA genes from the abundance table. Does that sound reasonable?

--
Sailfish is available at https://github.com/kingsfordgroup/sailfish
Citation:
Patro, Rob, Stephen M. Mount, and Carl Kingsford. "Sailfish enables alignment-free isoform quantification from RNA-seq reads using lightweight algorithms." Nature biotechnology 32.5 (2014): 462-464.
---
You received this message because you are subscribed to a topic in the Google Groups "Sailfish Users Group" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/sailfish-users/tCbqTV_ZZC8/unsubscribe.
To unsubscribe from this group and all its topics, send an email to sailfish-user...@googlegroups.com.
To post to this group, send email to sailfis...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/sailfish-users/e985d476-c356-4a4e-916f-5d9b5a4e190a%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Vasisht Tadigotla]

It also depends on which rRNAs are getting depleted. The 18S and 28S rRNA genes are repeats that are not annotated on the reference genome.  I’m not sure that it’s necessary to renormalize if the various samples are mapped against the same transcriptome. 

You received this message because you are subscribed to the Google Groups "Sailfish Users Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sailfish-user...@googlegroups.com.

To post to this group, send email to sailfis...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/sailfish-users/CAMEm53pJHBa-FER9_1K57YshGw3%3DuonirNcv6UqgRWtNJ3wyWw%40mail.gmail.com.

[Thomas Sandmann]

Good point! I am planning to add the transcript of the canonical rRNA repeat unit (from RefSeq) to the reference. That you "absorb" reads from all rRNA subunits.