I am finally getting around to transitioning our amplicon sequencing from MiSeq to NextSeq with UDIs to take advantage of the relatively new 600 cycle kits. Wondering if any of you have done this yet and have any recommendations, particularly to the following points:
1- What cluster density are you targeting?
2- What PhiX% are you using?
3- Where do others stand on the denoising with pooled quality scores and support in q2-dada? The data I have seen others generate on patterned flow cells appears to be highly suspect with respect to inflated richness. We will of course have controls on the runs, but I would be interested in what others are currently seeing on the 600 cycle kits specifically. Bonus respect points if you have knowledge from using zymo standards or defined communities.
Our density is in the range of 800-1000. We use PhiX at concentration of 5% for an amplicon library of 4 nM with loading concentration of 16 pM. We usually use the following kit: MiSeq 600 cycle v3 run (2 x 300 bp). However, it is different for a new NextSeq kit. For the new NextSeq 2000 P1 600 cycle reagent kit: 450 pM loading concentration and 40% PhiX.
Transitioning to NextSeq 600 cycle kits for amplicon sequencing, you should aim for an optimal cluster density that balances data output and quality. The PhiX% used can vary, but a common approach is a 20% PhiX spike-in. For denoising with pooled quality scores in q2-data, it's recommended to carefully check outputs, as the patterned flow cells can affect richness estimates. Always include controls in your runs to validate your data.
The Head of Sequencing at my old company got back to me and shared this:
I think we used the NextSeq 550
I do know we usually load 16S amplicon on the Miseq at 8pM rather than 12pM, so ~33% lower than typical high complexity library recommendations (I believe they recommend about 35% lower).
We also typically use ~20% PhiX, although as low as 10% has been used successfully, but higher quality data (%Q30) and higher % clusters passing filter (%PF) come with the increase in PhiX (but also, naturally, lower coverage).
Thanks all for the tips! Here is an update should someone find this post in the future. We are 3 runs in to using the P1 600 cycle XLEAP kits and have had great success averaging around 100,000 reads per sample (576 UDIs per run). Currently running 35% PhiX and a loading concentration of 1 nM getting ~80% passing filter and ~90% >Q30. There have been no issues with dada2/denoising and our zymo standards are accurately capturing the composition with a background measured in the 10s of reads, well below manufacturer spec. These kits are bringing the sequencing cost down to about 3$ a sample. We also have a library prep protocol that is about 2$ a sample (not including the initial synthesis of the UDIs). The protocol is here if anyone should be interested: