MiSeq Low Diversity - N's in primer solution

[esherman]

My lab does a lot of MiSeq sequencing of low-diversity amplicon libraries and we've been using a standard 10% PhiX spike-in with reasonable results. I just heard about the idea of adding a few N's immediately downstream of the sequencing primer landing site, such that the first few bases are very high-diversity and the software can easily discern clusters before the low-diversity hits. We're considering switching our low-diversity assays to this N-based system, but I have a question before I order oligos.

Does this solution only require that the first few bases of read 1 have N's (i.e. high diversity) or do the first few bases of every read need this high diversity sequence? I can't imagine Illumina would store cluster positions from the first cycles of read 1 rather than having the software re-call clusters for each read, but I just wanted to check to see if anyone has experience with this type of low-diversity workaround before using a flow cell to test out my hypothesis.

[Brian Bushnell]

It's ideal to not just put in some random bases, but also random bases of random length (perhaps 4, 5, or 6bp) to ensure color-balancing throughout the read. I'm not positive, but I believe both (random bases and random length) would also be beneficial for read 2.

[bilyl]

My lab does a lot of MiSeq sequencing of low-diversity amplicon libraries and we've been using a standard 10% PhiX spike-in with reasonable results. I just heard about the idea of adding a few N's immediately downstream of the sequencing primer landing site, such that the first few bases are very high-diversity and the software can easily discern clusters before the low-diversity hits. We're considering switching our low-diversity assays to this N-based system, but I have a question before I order oligos.

Does this solution only require that the first few bases of read 1 have N's (i.e. high diversity) or do the first few bases of every read need this high diversity sequence? I can't imagine Illumina would store cluster positions from the first cycles of read 1 rather than having the software re-call clusters for each read, but I just wanted to check to see if anyone has experience with this type of low-diversity workaround before using a flow cell to test out my hypothesis.

The Miseq is pretty robust with high diversity. I doubt you'll get that much better signal with playing around with Ns versus the time you put in.

FYI, the Hiseq has a new software update that deals well with low diversity samples with ~10% PhiX.

[pmiguel]

The Miseq is pretty robust with high diversity. I doubt you'll get that much better signal with playing around with Ns versus the time you put in.

I think you mean "The MiSeq is pretty robust with low diversity."
I agree. Maybe still an issue with getting maximum density with zero diversity samples. With 90% of the clusters in a single channel I think the "template" calling might be problematic at higher densities.

--
Phillip

[esherman]

Thanks for the input, everyone!

I did a quick test where I amplified a single region of the genome using PCR primers containing the Illumina sequences and 8 N's immediately downstream of the R1 sequencing primer. There were no N's downstream of the R2 sequencing primer and both an i7 and i5 index.

Other than some pretty severe under clustering, the run seemed to perform well and gave nice high Q-scores for all 4 reads. It looks like including N's immediately downstream of the R1 sequencing primer is all that's needed!

[nucacidhunter]

They have used similar approach in this paper:

http://www.microbiomejournal.com/content/2/1/6