Nanopore Improved Buffer Flow cells

TLDR: Nanopore have improved buffer flowcells in beta testing (I think releasing sometime in 2025) which reduce blocking and input requirements. We've tried them. They seem to work.

The longer story: Its that time of year (again) when Oxford Nanopore Technologies are holding a meeting and I can't be present at it. To everyone attending the NCM meeting in Boston, hello!

One of the advantages of not attending is being able to play with some new kit and flowcells. Pretty much whenever Nanopore get in touch and say "would you like to try this out" the answer is yes. On this occasion I am particulary interested. Over all the years now (terrifying!) that we've been using Nanopore technology for different applications we've found that flow cell performance can be reduced as a consequence of blocking - very long DNA becomming clogged in the pore. When Oxford Nanopore Technologies first released the wash kits which allowed you to recover throughput and effectively re-use a flow cell it had a dramatic impact on what we could achieve with individual flow cells. Nevertheless, this is still an issue. For example, we have done a lot of work (see https://www.biorxiv.org/content/10.1101/2024.08.16.608306v2) on optimising ultra long libraries (where the read N50 is in excess of 100kb) but we still need (relatively) a lot of DNA and we need to flush and reload the flow cells over the course of a run to get the optimum levels of throughput.

So where do these new flow cells come in? Well. As I understand it they have "improved buffer" (if anyone at NCM gets more info?!?!) and so they should not need to be flushed and reloaded as often. Ideally we might be able to achieve the same yield and read lengths from less DNA and (to the sound of applause from the amazing team who run the sequencers here at Nottingham) fewer loadings. This sounded too good to be true, so we set about testing it.

We've taken a variety of samples at a variety of lengths but for now I'm just going to talk about ultra long samples. We made sufficient library from a single sample of extracted DNA (using the excellent New England Biolabs monarch kits) from cells to load two flow cells side-by-side. The flow cells were approximately matched in pore count.

The libraries were looking pretty impressive (if I do say so myself) with 2,200 plus in strand and some very tasty N50s:

At this point we could just start to see the feintest hint of difference in performance of these two libraries. This is the same library split to run on two different flow cells and so as close as we could make it to identical:

To my eyes, this looked as though we were seeing more blocking on the original flow cell and that the improved buffer conditions were making a difference. To really see the difference, we need to look at the first remux:

This was when I realised these flowcells are doing something different and really quite cool. Running for longer and the difference becomes even clearer:

So what's happening here? We can look at the pore activity plots and really see:

At this point we could really see a difference.

So the super exciting thing is that we are getting more data over time from the flow cell. The problem with Nanopore (!) is that you can watch what is happening in real time during the run. Which sometimes leads to late nights. But when you see things like this:

The problem with checking things at odd times of the day is that one sometimes makes the wrong decision. In this case, we decided to flush and reload both the flowcells at the same time. I think (personally) we should have left the improved flow cell for longer - and we are kicking that off right now. But even so, the difference between the two flow cells continued.

My suspicion is that we could have hit close to the same total yield in just one load - but we will see this soon.

By the time the run was over we had 93 Gb of data from the original flow cell with an N50 of 111kb. The improved buffer flow cell ended up with 115 Gb with an N50 of 102 kb. Given these are the same library they should have more similar N50s and - to borrow a phrase from the nanopore world - mappign is in progress to have a look at that.

Overall I am very excited about these flow cells and what we can do with them in the future. A key feature that is needed please Oxford Nanopore Technologies is an automated pause on a run when the flow cell pore count drops to some user specified level. That will give me more sleep!