DESeq2 for targeted RNA-Seq

[aggp11]

Hi,

We plan to run some targeted whole transcriptome rna-seq using the AmpliSeq RNA methodology currently being sold by Life Technologies/Thermo Fisher. As a background on the technique, LT has designed 100-200bp probes for >20,800 features from RefSeq and provide that as a tool to study gene level differential expression. You can find more information here:

http://www.lifetechnologies.com/us/en/home/life-science/sequencing/next-generation-sequencing/ion-torrent-next-generation-sequencing-workflow/ion-torrent-next-generation-sequencing-select-targets/ampliseq-target-selection/ampliseq-transcriptome.html

I have been using DESeq2 for our whole transcriptome sequencing data and am wondering if DESeq2 will work for this targeted approach. I am reading/trying to understand more about the Math behind DESeq2 and on the surface it seems that DESeq2 should work fine because if I understand it correctly,"gene length" doesn't necessarily factor in the sizefactor based feature level normalization (unlike RPKM/FPKM, which in this case can be replaced by RPM/FPM). However, I am curious to know if I am overlooking something that might affect how DESeq2 would behave for such an experimental design.

I'll appreciate any insights on this.

Thank you!

[Wolfgang Huber]

Dear AGGP11,

there is no fundamental reason why DESeq2 should not work with data from such an approach. As you say, the crucial assumption is that the number of reads for a gene is (to good enough approximation) proportional to its abundance, and that the proportionality factor -whatever it is- does not change between samples. So I'd make some exploratory plots to check whether this seems to be true for the data at hand.

Wolfgang

[aggp11]

Dear Wolfgang,

Thank you for the response.

Praful