Friday, August 04, 2023

Apton (Super) Resolution to Be Acquired Fulfilled by PacBio

Wednesday night brought the news that Pacific Biosciences has both officially launched their Onso desktop short read sequencer and bought technology for a high throughput version of it by acquiring Apton Biosystems.  Apton had been developing their own short read chemistry and an instrument based on super0resolution imaging to go with it and was apparently relatively close to launch.  PacBio got Apton for $110M, with $85 paid now in PacBio stock and the remaining $25M "stock and cash add-ons" according to GenomeWeb (premium/free via ex-Twitter) - so presumable based on certain milestones being met.  
Super-resolution imaging is cool stuff, violating the rule laid down in intro physics class that it is impossible to resolve objects smaller than a particular function of the wavelength (the diffraction limit).  Such rule-breaking was rightfully rewarded a Nobel Prize in 2014 - though in Chemistry, so for once the chemists didn't get to whine about biologists horning in on their prize since it was physicists that time.  There isn't a true free lunch -- in general beating the diffraction limit means making tradeoffs or can only be achieved with certain constraints.  For example, you might be beating it in the X-Y plane but not at all in the Z axis. I'm not well versed in the subject, but one class of techniques involves imaging the same region multiple times and computational combining the results and/or illuminating the same region in different ways.

Illumina announced in 2020 super-resolution for the NextSeq 1000/2000 series to pack clusters tighter than before.  Tighter clusters means more per flowcell, so if the reagents are a major cost driver then that's more bases per reagent pack and therefore lower cost per base.  Plus who doesn't like more data?  The catch was slower runtimes, as the system used multiple images of each region to resolve the higher cluster density

Apton was going a similar route.  Some more details can be found in Nava Whiteford's 41J blog entries from June 2020 and September 2018.  Apparently they had both sequencing-by-synthesis chemistry and an instrument going.  PacBio claims the Onso sequencing-by-binding chemistry works out-of-the box on the instrument and still gives mostly Q40 results, which is impressive given the possible degradation of quality from super-resolution -- if you aren't careful signal may bleed from adjacent clusters or you may confuse tightly packed clusters.  PacBio does not appear to have disclosed the size of the instrument, but presumably it won't be a desktop, nor the cost nor launch schedule.  

If Apton was so close to having a marketable system, why did they sell? Presumably Apton looked at the remaining gaps in their commercialization plans -- manufacturing the instrument and consumables at scale, building out a sales and distribution network, etc -- and decided selling the company was attractive.  Perhaps even PacBio came a knocking.  But one can't help wonder if Apton looked at the market and saw how the new short read players are finding getting a grip in the desktop market challenging and decided that only an established player could possible put a dent in Illumina's monopoly.

If Apton did shop the company, it might soon be apparent that PacBio was the only realistic opportunity among existing players.  Illumina has their own super-resolution technology and would probably attract unwelcome regulatory interest in taking out an upstart short read player. Plus Illumina started the summer distracted by the huge unforced fiasco of Grail, then the proxy battle drama and is now without a CEO to set direction.   Apton's technology appears to rely on patterned flowcells, perhaps with DNA nanostructures involved, which is anathema to Element's team. Plus Element probably has their own high throughput plans well underway.   Singular Genomics has a market cap about one half what PacBio paid and only about 2.5 years of operating cash when they last reported in May; even if they could have afforded Apton the last thing Singular needs is the distraction of a major corporate integration.  Complete Genomics (aka  BGI aka MGI, or as somewhat asked at AGBT "that's what they're calling themselves this year?") already has ultra high throughput instruments.  Ultima is already in the ultra high throughput scheme and I'd guess (please correct me if wrong) that the spinning imaging of Ultima would greatly increase the difficulty of any super-resolution scheme.  Or put another way, Ultima isn't terribly dense now, so they can go an easier route of increasing density for quite a while before super-resolution seems like the way forward.

Of course, there could have been a big company interested in jumping in, but Agilent and QIAGEN seem to have decided selling sequencers isn't in their interest and Roche seems intent on buying up nanopore IP without ever launching a nanopore product.  Plus Roche has been focused on the clinical market, and an ultra-highthroughput instrument in that space is still questionable from a commercial viewpoint.

There's also the whole question of how big the high throughput market is.  Element and Singular are already blurring the line with their higher output kits aiming for the counting application markets -- or even Element's best data output with the standard kit or the longer read length versions that seem imminent -- 2x150 with Cloudbreak chemistry is about the same output as an S1 on NovaSeq 6000.  Many core labs or small biotechs, unless they are certain they will get huge projects, may prefer to buy a smaller, more space-efficient sequencer and just run multiple flowcells on occasion rather than invest in huge box that mostly sits idle or causes waits to efficiently fill flowcells.

Right before the SARS-CoV-2 pandemic took over, I did a lazy survey of sequencing startups.  I've tinkered with revisiting it, as it would be interesting to explore which startups are still in business -- so many have dead websites and there have been multiple on that list acquired (Apton and Omniome by PacBio; Stratos by Roche).

I should note that in general I'm bearish on the whole Onso concept; I'd much rather PacBio focus all their attention on their well-differentiated long read line.  Slides from PacBio are promising that development proceeds apace on both a benchtop HiFi sequencer and a monster high throughput instrument.  If the leviathan (I think I've seen a slide claiming 10K human genomes per year) significantly brings down the long read vs. short read price premium, that could really change things.  Plus PacBio should be seriously concerned with Oxford Nanopore's duplex technology eroding the quality difference in long reads.  The existing Onso has significantly less output than sequencers such as AVITI and PacBio's sales pitch that the higher accuracy is worth it seems to be running into significant skepticism in the marketplace.  PacBio really needs some heavyweight peer reviewed results showing a real benefit or most people will assume they've overshot the market demands for raw read accuracy.

Well, at least PacBio lived up the often slow summer news season.  Of course, slow is good if I want to clear my mental deck of a bunch of ideas that have rattled around.  This is a rare two-post week, can I make August a multi-post month?

1 comment:

Anonymous said...

Nice analysis from a business perspective, thanks Keith.