Loading Nextera XT in to MiSeq

[mdalessio]

Hello.

Just went to load my diluted DAL sample, but the MiSeq user manual wants you to add 10 ul of phiX DNA to 990 ul diluted sample.

The Nextera XT kit outputs 600 ul of diluted sample.

What?!

[microgirl123]

First of all, do you even want a phiX spike? If you do, then follow the Illumina instructions (from the MiSeq guide) to make a denatured 12.5 pM phiX library. Then decide what percentage phiX spike you want and do some math to figure out ~how much to add. For instance, if you want a 1% spike, you need to add 6 ul to your 600 ul library.

[mcnelson.phd]

If this is your first run, then I'd recommend adding phiX. This will help with troubleshooting if anything goes wrong. For example, if the run fails immediately because it can't find clusters, then you don't know for sure if it was your libraries or the system/reagents if you don't add any phiX which is almost guaranteed to form clusters (although not at a high enough density to save a run if only spiked in at 1%).

As microgirl said, just follow the manual for the standard denaturing procedure to dilute the phiX down to a reasonable level (I'd say 8pM, but to each their own). Then add 6ul of that to your 600ul of library and load into the cartridge.

Given that you're diluting phiX down following the standard method, this is one of the reason that I recommend doing the normal denaturing procedure because if you're doing the whole process for one thing you might as well do it for two.

[mdalessio]

Yes, this is my first run. I diluted phiX to 12.5 pM, and then added 5 ul of that, so far I am getting lower than ideal cluster density (~260K) but so far so good.

Makes sense about the denaturing protocol.

What bothers me is that the kit is made for use with this sequencer, yet there are incompatibilities between their protocols! I then also spent like 45 minutes this morning on the phone with Illumina because of bugs in the software.

[mcnelson.phd]

260K is low, and could be due to the bead normalization and denaturing protocol or simply that some of your libraries weren't that good and didn't form viable clusters. You'll get some idea after the indexing is complete but it's annoying not knowing for sure which is why the standard denaturing method is nice.

I'm not sure where you're implying that there are incompatibilities though? There are definitely inconsistencies, and I've noted my opinions with my FAS and area rep a few times about them. If you're talking about the XT protocol saying you dilute to a final volume of 600ul, but then the MiSeq manual saying you should add 5ul of phiX to 995ul of library, that's not really an incompatibility, just shoddy editing on the part of the manual writers.

My biggest gripe with Illumina really is how they make the MiSeq and Nextera library prep seem so simple that anyone can do it, but then omit to tell you the details that you need to know to get a good run. The best example of this is with cluster density and how that relates to what final dilution you use for your pooled libraries. When we first got out MiSeq the best our FAS could tell us was to go with 8pM for the final dilution and if it doesn't work then they'll replace the reagents. But who wants to waste an entire run to find out that they should have diluted to 12pM or 6pM instead of 8?

[mdalessio]

I guess inconsistencies isn't the right word, I was more referring to what you are saying about the perceived "simplicity". In their protocol they don't even explain cluster density at all. For the Nextera XT kit, they say, take 24 ul of the pooled sample and add it to 576 ul of buffer. Load this. No explanation about the IMPORTANCE of cluster density at all. No mention even of the concentrations, or how concentration relates to cluster density.

I still don't know how to determine if you should dilute to 8pM v. 12 pM v. 6 pM - explain?

[mcnelson.phd]

As I said before, this is probably Illumina's biggest failing, making their workflows/kits/systems seem overly simple.

Cluster density is probably the single most important factor affecting run quality, but they really don't explain it at all. I remember asking my FAS at the time why there was a table in the manual about how to dilute to various concentrations, but absolutely no explanation of how or why you would choose one or the other. She couldn't really respond because she agreed that they really need to get better at explaining those things, but unfortunately they still haven't.

As for determining the optimal final dilution, that's a bit of a black art I think. I've been working with the MiSeq for over a year and a half now, and while I've gotten quite good at determining what point I should dilute to it's still hit or miss some times. The general rule of thumb is that you dilute libraries with small fragment sizes more than libraries with larger fragment sizes.

For example, the 16S V4 amplicons are ~400bp, and we dilute to 6pM. This generally gives us clustering around 850K, so we could go higher but because of the high base skew we generally like to keep the density lower which also keeps quality higher. For most of our Nextera libraries, we generally get around an 800bp avg. fragment size, so we use 8pM for the final dilution. Because they're Nextera libraries with a wide distribution of fragment sizes, this usually gives us a cluster density around 800K, with some dipping down to 600K and others going up to 1050K.

If you're just doing a phiX control run, then Illumina says to dilute to 12.5pM, but they don't recommend that for actual libraries. Part of the reason is that while the system can support cluster densities up to 1250K, the quality, especially of read 2 drops a lot and you get lower pass filter rates. We'd rather have higher quality and pass filter rates than a greater number of poor quality reads, so we don't try to push cluster density that much.

[ScottC]

I completely agree about the Illumina gripes here. The MiSeq, especially in conjunction with the Nextera kits, was (and is) marketed as a very simple system that anyone can use, when really it's not that simple. We've been using Illumina systems since the first Genome Analyzer so it's not that big a deal for us in practice - we're used to Illumina's quirks (remember the V1 "kits" which came with powdered reagents that had to be dissolved, chemicals that had to be mixed, reagents that had to be purchased seperately?). It's a very thin veil between the bad-old-days of the GA and the MiSeq (although it's getting very slowly better). I've stopped reporting all the bugs, errors and inconsistencies to our Illumina contacts because they didn't seem all that interested... bigger fish to fry, so it seems. But enough of that rant!

As others have said, finding the right cluster density is pretty tricky, and Illumina doesn't really have a hard-and-fast rule about it because you'll see differences from library to library and from instrument to instrument. It sounds like a terrible way of doing it, but you'll get a feel for it once you've run your machine for a while (Back in the old days of the GA, they actually recommended that you run a titration flowcell for each library to get the cluster density correct... which, of course, was ridiculous, since that would have cost perhaps $10K!)

Another frustrating thing (in my opinion) is that Illumina's technical information
and instructional information is quite fragmented... there's a lot of information that is spread across may different publications and PDF documents, and you just have to know where to look. Don't hesitate to contact your FAS and bug them for information about anything you need to know.

A low cluster density could also be an issue with denaturation. Make sure you do it 'by the book' - heat it up and cool it as quickly as you can. If you're using NaOH (probably not with NXT), then make sure you use fresh NaOH.

[mdalessio]

Thanks for all of the comments.

I agree that the information is very fragmented. I think I am going to try to load ~8 pM and try again.

[gwilkie]

We gave up with the bead normalisation method for Nextera as it was very inconsistent in our hands.

Instead, we calculate the molarity of each library using a Qubit and Bioanalyser before pooling samples and spiking PhiX at 1%. We then check the final 4nM pool for size and concentration.

We use the NaOH denaturing method (fresh NaOH every time) and load at 12.5pM for MiSeq v2 chemistry. This works well and consistently gives us a cluster density in the 950-1100 K/mm2 range.