Wednesday, May 21, 2025

Oxford Nanopore Should Spin Out Protein Sequencing

I've toyed with writing something on these lines for a long time but never quite pulled the trigger.  But the more I think about it, the more imperative my logic feels for spinning off the nascent protein sequencing effort.  I actually finally peeked at the agenda for the meeting and University of Washington's Jeff Nivala is basically closing the meeting with an update on his work in this space.  

It is my opinion that Oxford Nanopore has often had parts which could be separated from the company and that such a separation would benefit the technology being pulled out  in terms of probability of success.  I'll first walk through the idea with DNA synthesis, DNA-based data storage and VolTRAX and then finish up with protein sequencing - as those other three are dead branches of the ONT effort.

The central thesis is one of focus, and I will upfront admit that I am an expert on the dangers of lack-of-focus due to often falling into loss-of-focus traps - and by falling in I really mean charging head first eyes wide open.  

We can divide the aspects of focus for a company into several parts.  First, there is the allocation of resources - particularly human talent.  Second, there is the body of key opinion leaders (KOLs) and potential customers.  Third, there are the monetary aspects.  These aren't wholly separable, but it is useful I believe to at least pull out these foci when thinking about this.

Resource allocation and the monetary aspects have one key coupling: what incentives should a company be responding to.  When ONT was younger and private, their main incentive was to generate buzz that would draw more attention to their platform, which ideally would bring in KOLs and customers.  But ONT is a maturing company with a focus on revenue generation, and that is a very different beast.  Any long term technology development project is by its nature not going to deliver revenue and financial growth for many years.  In a public company, management must focus on keeping investors happy - the alternative is for an activist investor to march in and make radical changes (a misery Illumina was subjected to).  

For KOLs and customers, many of these spaces have weak overlaps.  You'll always find some biological polymath or an unfocused company which will cross them, but in general scientists performing large scale genomic studies are a different population than those performing large scale DNA synthesis who are in turn different from those interesting in committing to exploring DNA-based data storage - and very few individuals from any of these groups are going to be interested in protein sequencing as it is likely to be realized by ONT.

The counter-argument is one of synergies and avoiding friction.  It would be necessary, for example, for a nanopore protein sequencing company to rely heavily on intellectual property of the DNA sequencing company, and that might include chip designs and software infrastructure.  But, there are ways to solve that through licensing.  I'll also say that nearly every company I've worked for (Infinity is the notable exception) had some concept of disparate parts trying to leverage synergy by staying together, and in retrospect none of them managed it very well.  Of course, I don't have access to a parallel universe in which the decision to break up was made - other than at the strain factory which last summer did break into much more loosely coupled individual business units - an event that certainly has strongly sharpened my daily focus.

ONT had a tendency to trace hot trends.  When alternatives to phosphoroamidite- and column-based synthesis late in the previous decade, ONT eagerly revealed they had hopped on the bandwagon.  In a London Calling talk, Clive Brown gave few details but vowed that this "DNA Foundry" unit would have its own signature conference and exciting results would be revealed.  No such conference ever launched and the effort was soon absent from any presentation.  What technology angle ONT was up to was never revealed, though rumors went to using the VolTRAX electrowetting technology in some form.

A number of new DNA synthesis companies are now hitting the landscape with product offerings.  Most of these are aiming for the gene synthesis market - longer DNAs made faster and with fewer design constraints.  That's a field that will have a nearly completely different customer base from genome sequencing.  There have been a few companies with "make oligos on your benchtop" offerings and perhaps that's what ONT was aiming for.  But that's a very risky proposition, given the large companies that can deliver oligos overnight for very low prices.  There's reasons to still like having local production, but it's still tough competing with existing vendors with extremely good service offerings.  

Similarly, when DNA storage was becoming hot, Clive took some of his London Calling presentation to describe a proposal for doing this with native DNA.  I absolutely hated it.  Curiously, one comment on my Clive Brown piece suggests I've cloaked this opinion out of embarrassment that there are publications describing similar approaches.  If so, I've failed - I thought I made it clear I still hate it as an ONT effort.  Great from an intellectual standpoint, fun to think of a latter-day George Smiley swabbing drop points for steganographic messages to read with a MinION, but from a commercial point-of-view I still think it's an awful idea.  Clive proposed using the tech to archive personal items like photos and music.  There's definitely a market for that - but if I want to do that do I (a) try out some strange new technology that requires refrigerated kits to store a few hundred gigabases that can be read slowly - maybe - or (b) order a 1Tb industry standard drive for less money?  If you answered (a), you are in the 0.001% of the 0.001% of the human population.  And if I'm wrong and the tech could be made to work absolutely reliably and did have a market, it was a long ways off.  Far better to have that tech talent focused on improving the core platform which was starting to generate revenue.

VolTRAX is a related but slightly different argument, though we know now that VolTRAX has performed so badly that ONT has given up on it.  But teleport back to when excitement was high.  The VolTRAX vision was highly compatible with the overall ONT vision: a low capital cost, portable sample and library prep device that would be easy-to-use.  But having such a device be limited, by corporate design, to only service the ONT community would be a strong restriction on the product line's potential revenue.  Separate from ONT, such a device could be marketed to users of other platforms, or perhaps even for non-sequencing applications.  A bigger potential market means potentially enticing larger outside investment which could then maximize the chance of technical success - not every technical roadblock can be vaporized by shooting money at it, but it is a method that works with significant frequency.

Okay, so now back to protein sequencing.  It's in its very early stages and the company is starting to see real pressure to perform financially.  If nothing else, it is likely to result in management creating high-level goals to present to the board and the investment world that don't have protein sequencing on them, or it is at best an obvious hanger-on.  

Looking at the other aspects underscores that protein sequencing is a different beast than genome sequencing.  There are crossovers - the Olink's and SomaLogic's of the world trying to read protein levels much as sequencing reads RNA levels.  But the proteome has some big technical challenges.  Foremost, interesting biological samples have a mind-boggling dynamic range.  If you are counting peptides, then you must cope with the fact that in something like blood most of the peptides are from really dull stuff like protein albumin and beta-2-microglobulin and most of the interesting hormones, cytokines and the like are 8, 10 or maybe even 12 logs below that.  So unless you have extensive workup to deplete the boring or fractionate the sample, or you have affinity reagents to select for the interesting stuff, no near term nanopore workflow will be able to analyze blood, plasma or serum in an interesting way.

Everybody in the space copes with this.  QuantumSI is the only "next gen protein sequencing" company that has a launched product, and they've been struggling to find markets.  It also doesn't help that there is established technology - mass spectrometry - for protein readout which is good.  It has flaws - the kit is extremely bulky, expensive and demanding equipment requiring specialists.  A number of those specialists are obnoxiously cocky of their position, which doesn't help.  

Then there's a growing number of affinity reagent based companies that are setting standards for cost and sensitivy - I mentioned OLink (now part of ThermoFisher) and SomaLogic (owned by Standard Biotools and partnered with Illumina), but then there's Alamar and Nomic and Codetta and I think yet another just emailed me.  A key limit on these approaches is needing the right affinity reagents - you're generally in luck if you work on human or mouse but out-of-luck for just about any other species.

I think the nanopore proteomics idea is very interesting and I'm looking forward to hearing Nivala's talk, but seriously worry that in an increasingly (and not by choice) bottom-line focused organization it won't get the proper backing.  If separate, it could also draw on new investment sources.  And the community of scientists interested in some sort of protein characterization at the scale likely to first emerge - well, that won't draw many from the current London Calling attendee crowd.

In a dream world - and for ONT fans this should be an attractive dream - ONT would evolve into an ecosystem resembling that of the electronics world - the supply chain is broken into multiple companies that focus on very specific pieces of the business (though it is also true that they are increasingly trying to poach from each other). So TSMC focuses on chip making and not designing, NVIDIA on chip designing but not end products and Apple on end products.  If such a constellation of nanopore companies drove the technology forward at high speed and brought in tremendous outside investment (which would enable that rapid technological trajectory), that would be a win for the space as a whole.


3 comments:

  1. Essentially what Portal Biotech is?

    ReplyDelete
  2. Any thoughts about the ONT based proteomics now that the data is out? Seems like you can get very high sensitivity but you need to know what you're looking for, and have an antibody against it..

    ReplyDelete
  3. When OnT has spun things out for exactly the reasons you outline it has been hard to find optimal dedicated leadership in the UK and they have been spun back in again and then frozen.

    ReplyDelete