Novaseq vs Miseq failure

[Tcom111]

Hi all,

This is the first time I'm trying to run a MIP (molecular inversion probe) library of ~800 genomic targets, dual index library in Novaseq.
The primers we use are custom primers for all reads (R1, R2, Ind). The Miseq run of the same pool looked like this (Miseq1, Miseq2 figures), while there was a fall in the first 50 bases of both R1 and R2 in the Novaseq run (Novaseq1, Novaseq2). Has anyone had any encounter with such phenomena and can suggest how to solve it? I heard an explanation of low hybridization efficiency of the custom primers in the Novaseq, but how can this explain the rest of the sequencing process?

Also, I'll be glad to hear from satisfied/non-satisfied Novaseq users general experience compared to other machines (namely Hiseq).

Many thanks!

[pmiguel]

The new version of SAV has a metric "% occupancy" for each tile. You should check that.
It looks like you either had very low occupancy (very under-clustered) or very massive over-clustering. (Actually I think the NovaSeq is pretty resistant to over-clustering.)

I personally would be terrified to use custom primers on an S4 flowcell. Actually we never used them even once on our previous instrument (HiSeq 2500).

--
Phillip

[GenoMax]

MiSeq is the champ of sequencing difficult libraries. I would not compare success on a MiSeq to a different sequencer for odd libraries. If you are running an un-supported application then you are on your own when you push the bounds of technology.

When you stay within bounds of supported applications NovaSeq performs very well.

[mama]

Which version of SAV enables to see the % occupancy per tile. How can you check if the loading concentration looks good?
Many thanks!

[Tcom111]

Thanks for your replies!

I personally would be terrified to use custom primers on an S4 flowcell. Actually we never used them even once on our previous instrument (HiSeq 2500).

and

When you stay within bounds of supported applications NovaSeq performs very well.

Novaseq and other machines have a specific protocol for custom primers, although I understand the fear.. I believe that it may have to do with the machine different temperature differences.

I was told that the intensities are low, which may resulted in faulty results.
Any thoughts about that?

[pmiguel]

Originally Posted by mama

Which version of SAV enables to see the % occupancy per tile. How can you check if the loading concentration looks good?
Many thanks!

SAV 2.4.5 includes an "occupied count" and "% occupied" metric in the drop down list of the Flow Cell Chart pane of the Analysis tab.

--
Phillip

[GW_OK]

One thing worth keeping in mind as well is the difference in clustering chemistries between the two instruments. With the Nova using ExAmp and the Miseq using the "classic" bridge amplification the same library may (and probably will) cluster differently between them. We were actually warned by our FAS during our Nova install to not QC pools on a Miseq Nano prior to a large Nova flowcell and instead run on an iSeq since it also uses ExAmp.

I'm wondering if your poor Nova run was due to presence of adapter/primer dimer, which can easily take over with ExAmp. Do you have an electropherogram of the library and/or FastQC results showing adapter dimer or insert size metrics?

[Tcom111]

Hi all,

As a followup on this issue (I owe it to the generous people here).

I ran the same library in Miniseq which has similar temperature properties and it looked really good. As a final step, I noticed the sequencing provider had some technical issues, including not supplying coherent answers as for what they did, and moreover, were unclear of what they suggest to do (at the end, they said they will rerun with addition of Phix and mix standard and custom primers). I chose to run in a different facility which supplies excellent support even before sending samples. The run had great QC, similar to the Miseq data (!) with S2 chip, yielding 3.7B reads.

Bottom line, Novaseq can do custom primers. And sometimes you should be lucky when selecting the provider (both sequencing provider have reputation and certificates etc.).

See attached figures for the successful Novaseq run (see posts before to compare).

Best.

[GenoMax]

I am glad things worked to your satisfaction. Thanks for providing a final update.