Anyone using NovaSeq 2x250 for 16S sequencing?? 1 billion + reads..

[samd]

Title says it all.

This seems like the future for 16S sequencing but haven't heard of anyone validating yet. Seems like 2x250 is enough coverage for 16S and 1 billions plus reads is WAY more cost effective and efficient than Miseq.

Cheers,
Sam

[itstrieu]

It seems like over kill unless you have enough samples/index to fill out the flow cell. We usually sequence @ 100k reads per 16s sample. I can see it being more cost effective if you are doing metagenomic shotgun sequencing on the NovaSeq compared to the NextSeq.

[samd]

Yeah of course for small libraries, the Miseq is enough. But we and a lot of other labs are moving towards experiments with huge sample sizes (300+ to even thousands of samples). Having to do multiple Miseq runs for such libraries is a pain in the ass because of run bias. You can completely wash out an experimental result with run bias alone. One clean NovaSeq run would be much better from both a cost and experimental perspective.

[itstrieu]

It makes sense using one NovaSeq kit compared to multiple MiSeq kit at that volume. Our break even point is around ~1175 samples on the NovaSeq and TATs becomes an issue for us at that point.

If I am validating the NovaSeq for 16s sequencing, I would add diversity to the libraries to ensure all 4 DNA bases represented in every cycle (spacer primers, more PhiX, etc.) since it uses a two color chemistry. Also the NovaSeq uses a pattern flow cell which is more sensitive to adapter dimers so I would make sure to clean the libraries well.

[samd]

Interesting. I was not aware of these issues that are unique to NovaSeq.
It sounds like there is a timely need for a Miseq NovaSeq comparison........

[itstrieu]

Oh yea there's definitely a need! Also depending on how sensitive the experiment is, there is also the issue with index hopping. But I see this only as an issue if you are looking for rare variants.

[samd]

Yeah very true. We actually moved over to Ilumina's unique indexes to deal with just that issue. I also know labs that use "staggered" primers which fix that issue as well while also increasing diversity so you don't need PhiX anymore.

Just brainstorming here..but all you would need to do is send the same library to both a Miseq and Novaseq to compare the outputs? And I don't think you would want to control for sequencing depth since that would essentially be the point of NovaSeq anyway.

IDEALLY you would want a controlled comparison where the reads per sample are the same and an uncontrolled comparison where the sample numbers are the same and observe the results of the two comparisons.

[itstrieu]

I am not an informatics person but maybe you could normalize the reads by downsampling. I would run past libraries on the NovaSeq or prep new ones to run on both platforms and compare. It might be helpful to run some standards like mock microbial communities to limit any basis you might get from extraction methods.