Amplicon library loading concentration on NextSeq 2000

[ajthomas]

I've been sequencing amplicons (mostly 16S, but a few others as well) on a MiSeq for many years. Some of my users are migrating to the NextSeq 2000 for their amplicon libraries, but I'm finding it a bit more fickle than the MiSeq. I have found the loading concentration generally needs to be quite low, but the optimal loading concentration is still quite variable. I have had one 16S library that sequenced well at 600 pM, and others that are overclustered at 100 pM. I now generally load amplicons at 100 pM, but my pass filter rate with 100 pM varies from 12-90%. To check my quantification, I quantified some of these libraries using Qubit/TapeStation and qPCR and the numbers agree pretty well.

Has anyone else seen such unpredictable behavior from amplicon libraries on the NextSeq 1000/2000? How have you dealt with it?

[UzoSeq]

If your original library is highly concentrated (> 20 nM), I suggest that you dilute it (1:4 to 1:10) before taking the amount you need to prepare for the sequencing experiment. The reason is that you risk pipetting wrongly when the concentration of your original library is too high, as a little as 0.5 - 1ul pippetting error could introduce far more amounts than you had calculated. Next, spike with 20 - 25 % of PhiX, I hope you will have a consistent clustering this time. Good luck.

[ajthomas]

I always dilute down to 2 nM first then requantify to double check before proceeding with diluting down to the final loading concentration. I also use plenty of PhiX. I've been running Illumina libraries for many years now and thought I had it down. This new instrument, however, has been quite finicky with amplicon libraries. On the MiSeq and NextSeq 500, the optimal loading concentration for different library types was fairly similar, and generally consistent from one library to another. Not so with the NextSeq 2000, with amplicons needing loading concentrations about 10-20% that of other library types, and being much more variable. Have you sequenced amplicons such as 16S with your NextSeq 2000? If so what has been your experience?

[UzoSeq]

I am sorry, I do not have experience with NexTSeq 1000/2000, hence I was unaware of its abnormal behaviour.

[clamoris]

Yes, you're definitely not alone — amplicon libraries, especially 16S, can behave unpredictably on the NextSeq 2000. We've transitioned a few workflows from MiSeq to NextSeq 2000 as well and noticed similar inconsistencies. A few key things we’ve observed:

1. Overclustering is a major issue with dense, low-diversity libraries like 16S. Even at 80–100 pM, we've seen poor pass filter rates depending on the exact library prep and complexity.
2. Low-diversity compensation: Unlike the MiSeq, the NextSeq 2000 is much more sensitive to low-diversity issues. Make sure you’re spiking in a higher percentage of PhiX (we’ve gone up to 25–30% for some 16S runs) to help balance the base calling.
3. Library prep variability: Even when Qubit/TapeStation/qPCR agree, subtle differences in fragment size distribution or adapter ligation efficiency can impact clustering efficiency. We’ve started normalizing based on qPCR only and size-selecting more strictly to get more consistent clustering.
4. Loading concentration sweet spot: For us, 80–120 pM seems safest, but we always test a new prep at multiple concentrations first to calibrate. Once dialed in, repeat preps tend to behave more predictably.
5. Try XP workflows: If you're not already using the Xp workflow for loading (especially for 100-cycle kits), consider switching. It gives more control and reduces clustering variability compared to P2 loading.

[zkartje]

Hello ajthomas,
For amplicon-seq, we have recently started sequencing on NextSeq 2000 and noticed a high frequency of SNPs arising within the amplicon that are not present when sequenced on the MiSeq (70-80% prevalence). These are not present whatsoever on the MiSeq, making me think these are an artifact of the sequencer. We routinely use 20%+ PhiX to compensate, but it still seems to be an issue. The NextSeq cartridges are much more cost-effective than the MiSeq, but these issues make it difficult to completely transition.
I have only tried 500-600 pM loads so far. I will try spiking in even more PhiX and loading less to see if that helps.

[Gmurray90]

Hello, i work within a Sequencing Facility. How concentrated are your amplicons at the QC step? We typically dilute down to between 2-10nM prior to qPCR to try and reduce dilution errors that may occur from highly concentrated samples when sequencing. We have a NextSeq 1000 and load 650pM with 40% PhiX (or 1% plus another library type to increase complexity). We typically get >75%pf.