Genomic DNA libraries using Nextera

[creeves]

I am still trying to wrap my head around using Nextera kits to get the maximum quantity and quality of sequence data from a MiSeq loaded with a pool of three yeast genomic DNA libraries (12 Mb). We have obtained acceptable data after three runs, but it has been hard to control the size of the library fragments.

Is it true that once one of the transposomes in a tagmentation reaction inserts its adapters into the target DNA, the unloaded transposase cannot cause any further mischief? If so, then knowing the number of active transposomes in the 5 ul of TDE1 or ATM should allow one to calculate the amount of DNA of a given size to combine with the 5 ul TDE1 or ATM to achieve the desired size distribution once the reaction has gone to completion. It is after all a poisson process.

Allowing tagmentation to go to completion would get around the variable results of a timed reaction. What is the flaw in this logic?

[User001]

Your logic sounds good. We are also using Nextera but have not tried to change the tagmentation reaction time.

[mcnelson.phd]

Overall your logic seems fine to me as far as the activity of the transposes is concerned but there are some finer points that I think are incorrect.

First, the transposomes don't target the DNA in a truly random process. They seem to do a pretty good job at being as close to random as possible, but the fact remains that any transposon will have a site bias for where it will prefer to bind and act. This means that your input can have a large effect on size distribution that simply trying to tagment more won't overcome. We've seen that some bacteria with "weird" genomes and a couple of plastids have consistently given us different size distributions than what we see for most normal genomes.

Second, Epicentre/Illumina supposedly already did all of the titration and timing to determine how to achieve the preferred size distribution while also ensuring that a viable library can be created. We've tried playing around with the incubation time to see if we can shift our modal fragment size either higher or lower depending on what we want and extending the time up to ten minutes didn't have any appreciable effect on size. So, the 5 minutes that Illumina recommends may actually be pretty close to that end point for most libraries if you start with the recommended amount of DNA.

Saying all of that though, you're certainly welcome to play around with the incubation times yourself and report back to all of us with what you find. One of the greatest things about this forum is that most of us love to tinker with these types of things and hopefully we can all benefit from each other's failures and successes.

[epistatic]

We have been struggling with the Nextera exome prep. We definitely input 50 ng of DNA (picogreen and Qubit) as directed and all of our samples came out to a 2 kb mean fragment size. It was reproducible but not close to what I want for exome capture.

We tried several things to optimize and had many calls with Illumina. What we found out is that the number of active transposomes in the reaction is the most important and the reaction goes to completion so adding more time didn't help.

With the lot of transposase we had the only way to get smaller size was to run a titration of input DNA. The 10 ng was too small but 20 and 30 ng input gave a more reasonable size range while 40 and 50 ng both gave a mean >1 kb.

Illumina tech support told me that every lot of this could have different activity so I would need to do an input titration to get the optimal fragmentation size. We are waiting to get data to see if lowering our input to 25 ng instead of 50 ng impacts our library complexity and exome capture stats.

After all of this I think I prefer the consistency of the Covaris, even if the Nextera prep could be easier. Multiple people at Illumina did tell me there was a trade-off in data quality using Nextera. Ultimately good data is what we need and I'm willing to put in the time to make good exome libraries. I think we will use Nextera for quick genomes and avoid it for exomes.