Custom SAGE with RNA SEQ Kit

[moritzhess]

Hi,
I posted this in the library prep forum but I assume it might fit in here better:

I want to use the Illumina TruSEq Kit for SAGE analysis: getting only one read per transcript. To achieve this, I want to fragment the RNA (or prior enriched mRNA) and then perform another rounds of polyA enrichment in order to only get the polyA streches with app. 100-200 bp of UTR region. cDNA synth. will be performed using polyT primers instead of the random ones.
I included a very ugly cartoon of the method as attachment.
Has anybody of you tried a similar protocol, am I missing some pitfalls?

In keen anticipation

Moritz

[genlyai]

Hi Moritz,

Can Illumina's DpnII-based digital gene expression methods get you where you want to be? The issue I see with your idea is that even if you use an "anchored" oligo-dT primer (TTTTTVN), you will still have to start your sequence either with a run of Ts (which will kill the sequencing run under normal circumstances, and will at least waste bases), or you will be sequencing a semi-random sequence from ~200bp upstream, rather than a defined tag.

I guess I should ask what advantages you're hoping to get with your protocol.

[moritzhess]

Hi Genlyai,

thanks for your response ! I heard that Illumina disconitued the DGE Kits. Furthermore as far as I know, the tags are only 20bp long, which could be a problem, because we don't have a genomic reference availalble. For the random character: I would not mind that too much. I intended to align the sequences to our reference transcriptome and then count the hits per reference transcript. But this TT problem might really be an issue. I hoped that polyT priming would be sufficient to get rid of the polyA stuff. Another option would have been to perform parired-end sequencing, My intention to use the protocol was to avoid high costs associated with sample prep, because we will have plenty of samples to analyse and the only available Kit for DGE profiling (3DGE ovation from NUGEN) will cost about 200 € per sample compared to about 80-100 € when using TRuseq. Furhtermore it is optimized for low RNA quantities whicht might affect the transcript diversity. There has been a method published "3SEQ" http://www.plosone.org/article/info%...l.pone.0008768 Which circumvents the polyA issue by using custom primers. But I am not that biotech genius and intended to use something that is more "low-tech" . Furthermore as far as I know no other publications has cited that method but maybe I should get in contact with that group.

[genlyai]

Yeah, if the kits are discontinued, then I could see that would be a problem. The paper you cited seems fairly straightforward ... it is just an adapter synthesis away from what you're proposing, and allows you to preserve directionality.

However, I'm confused here. You say you have no reference genome, but you do have a reliable transcriptome assembly? If so, it is probably easiest to just do mRNA seq. If no transcriptome, then I would caution against the randomly placed tags both you and the paper use. Most transcripts are low abundance, and that means you will have a hard time assembling most of the tags you get back.

One could imagine using the above paper with a restriction enzyme digestion in place of the fragmentation and get rid of this problem, but again: why not just mRNA seq?

[moritzhess]

It's good to hear that the method cited is not too complicated because I m really tempted to use their technique. We have a reference transcriptome. If I got you right, many transcripts in our samples may not be included into the reference, so they would have to be assembled from the unaligned reads ? We were frequently discussing about using full length RNA-sEQ or Tag-SEQ and we desided (not definetly) to use tag seq because of money issues. Using tag-seq we can use single end sequencing and we would not need such a high sequencing depth. Furthermore I expected the analysis to be more easy.Another benefit of tag-seq would be that it is not so sensitive regarding RNA quality. But your post makes me thinking again about RNA-SEQ . Do you have any recommendations regarding the sequencing depth using full-length RNA-Seq ?