Called Back To London Again

After a too long pandemic-induced hiatus, I'm in the UK for this year's edition of London Calling.  I talked myself out of going last year well in advance, which would have been interesting as my rapid tests were still coming up positive about the time I would have needed to fly from Boston over the Atlantic.  And while I've been watching remotely, I've been dismal over the past year in actually writing anything about it.  Which was foolish on my part as ONT has been going through an interesting transition.

What's been happening over the last year is a huge corporate pivot as Oxford standardizes on Kit 14 chemistry. Around year's end, nothing else should be available.  This will greatly simplify Oxford's product line offering, which should have many benefits -- a significant degree less in new user confusion and also a lesser challenge for ONT to ship the correct products out when you order them (let's just say I've seen the contrary happen too often).  While this pivot is happening, ONT has been surprisingly focused on this and the amount of other platform changes have been modest -- though certainly there are others and they are interesting too.

A key advantage of Kit14 over prior chemistries is generally higher accuracy and increasingly bold claims of yield for duplex sequencing.  There are two interesting twists on the accuracy front.  

First, ONT had originally said a year ago that the best accuracy would require substantially slowing the nanopore translocation speeds, which inherently reduces yield.  So the message was: choose high accuracy or high throughput.   Then at December's Community Meeting they issued a "never mind" -- with more data to train their models, the 400+ bases per second translocation speeds can generate similar high accuracy data.

Second, ONT demonstrated some of the foresight in designing the platform and how different changes over the year are now converging to enable even better performance.  Their devices had used 4Khz sampling rates, but the underlying hardware can be goosed for higher rates, so soon 5Khz will be the standard.  With higher sampling rates, ONT is confident they can deliver higher accuracy basecalling with models trained on these rates.  But it shouldn't be forgotten that a key step in enabling higher sampling -- which means more data -- is the the switch from FAST5 to POD5 format which is not only more compact but better able to handle large amounts of data.

It's worth noting that if you had an early MinION, it will still work -- as would a MiSeq you purchased at the same time.  But the MinION needs nearly no maintenance whereas MiSeqs do -- and perhaps more importantly the performance specs for the MinION keep climbing but the MiSeq hasn't seen any updates or new kits in a very long time.  A new version of the MinION is coming out, with sleeker styling, better temperature control and the ability to drop into a special dock to enable running from an iPad -- but you won't be forced to retire your classic MinIONs. 

I suspect a lot of focus at this meeting will be on duplexing rates and yields.  ONT has been offering to their select developers "high duplex" flowcells; presumably the meeting will be used as a forum to announce greater availability.  ONT has been secretive about how these flowcells enable higher duplex rates   It will also be interesting to see how ONT will be supporting customer demand for both high and low duplexing -- duplex accuracy is great but inherently reduces yield and for certain applications (particularly counting applications) yield is far more important.

There will certainly be much more.  ONT has shown a prototype device for reading cell-free DNA straight from blood.  It's a fun concept, but whether it will achieve the necessary sensitivity to compete in this space remains to be seen -- or if the marketplace and regulators are ready for such a device in point-of-care settings.  It wouldn't be surprising if ONT is feeling sufficiently confident in the Kit14 pivot to start expanding the product line again with the long-promised new ASICs to support tiny (SmidION) or arrayed sequencing elements (Plongle).

So stay tuned -- I promise that I'll actually write this meeting up!