polyA sequencing

[linjc]

Hi, I hope to do a polyA enriched RNA-seq on Hiseq2500, which only sequence 3' tags with polyA tails. I realize the polyA tail will affect the sequence result. Usually, we mix libraries with PhiX to enhance the diversity of nucleotides. However, it waste a lot of reads. Do anyone have any experience or suggestion on this?
Thanks.

[Brian Bushnell]

I won't claim to be an authority on this, but... low diversity seems to be problematic for Illumina base-callers, in my experience. Illumina has often changed its guidance on recommended PhiX concentration. I'm not convinced that they do this on the basis of actual science; rather, I suspect it's more based on their marketing department. My observations are mostly based on MiSeq, which has had catastrophic quality problems from low-diversity libraries in the last few years. I have not personally observed such issues on HiSeq (to which we currently add zero PhiX). However, we only sequence low-diversity (amplicon) libraries on MiSeq.

[nucacidhunter]

one approach would be using custom sequencing primers that covers polyA so sequencing starts from base following polyA.

[linjc]

if this works, that would be great. Have you ever try it?

[nucacidhunter]

I would suggest looking at the following link for ideas on this approach if you have already prepared libraries. You may have to reduce few bases from 5’ end of R1 sequencing primer depending on the number of polyA bases to adjust primer Tm and also would not be able to use PhiX as using both custom and Illumina primers will interfere with binding and sequencing most likely will fail.

If you have not prepared libraries it is possible to switch location of P5 and P7 adapters so sequencing will start 3’ to polyA site. Also you can design adapters with does not share sequence with Illumina R1 (or R2) primer binding motif to be able to add PhiX or multiplex sequence those with standard Illumina libraries using both standard and custom sequencing primers.

https://www.lexogen.com/quantseq-3mrna-sequencing/#quantseqworkflow