Thursday, September 24, 2015

Farewell Nabsys

A bit over a week ago brought news that mapping instrument hopeful Nabsys had ceased operations.  As a veteran of one failed biotech, I have a lot of sympathy for the team there. Plus, I knew a bunch of folks at the Providence RI firm.  Nabsys's signle molecule mapping technology was a wonder -- what single molecule technology isn't? Already stories are emerging of a disgruntled founder who wants to buy up the intellectual property and give it another go. It is easy to admire that stick-to-it spirit; it's a lot harder to find a rational reason to believe that such a revival will be any more successful.

Nabsys had developed an all-electronic single molecule physical mapping approach, though unfortunately many news reports mistaken referred to this as DNA sequencing..  Mapping isn't a crowded field, with two direct competitors: OpGen and BioNano Genomics.  The idea in general is to identify the spacing between landmarks on long DNA fragments.  Make enough measurements sufficiently accurately and the patterns can be aligned to build a map of the genome of interest. 

Physical mapping has long played a role in the genomics, frequently serving as an important means of scaffolding and verifying sequence assemblies. Nabsys' specific technology involved converting DNA to long single strands and hybridizing probes to the DNA to mark certain regions.  With their nanochannel instruments, single stranded and double stranded regions (boosted by further coating the DNA with protein to both make it stiffer and bulkier) yielded distinctive signals separated in time.  Once the signals were corrected for the effects of hydrodynamic drag  (A video recording of a seminar earlier this year is my source for the details).  Because Nabsys used an electronic detection scheme and DNA in solution, their instrument could be quite small -- the heart of the system were somewhat egg-shaped units you could hold in your hand, though Nabsys placed these in a larger liquid handler.  In contrast, OpGen's optical mapping system was a huge instrument to support their scheme for stretching DNA, and BioNano's is a large desktop instrument.

I had visited Nabsys back in the early days of my current gig, with the hope of establishing a collaboration around our need to assemble large numbers of very difficult genomes.  Of particular interest would be solving the repeat problem: in most bacterial genomes large sequence repeats are often of secondary interest, but because with Streptomycetes a major class of large repeats drive secondary metabolism, they are of primary interest.  Nabsys was located in a beautiful building adrift in a sea of parking lots; Providence has some really nice areas but Nabsys was a bit outside of them.  We had a great meeting (enhanced by a very friendly spaniel in the conference room) and hoped to collaborate once Nabsys got along a bit further technically.  

We never did quite collaborate, for the two reasons that I think ended up dooming Nabsys. Bacterial scaffolding/assembly was a great starter market for Nabsys, requiring smaller amounts of data and simpler algorithms, and the area has been booming in interest as various public agencies have grasped both the breadth of bacterial diversity and the importance of high resolution genomes.  

However, mapping will always be less satisfying than sequencing, with a highly fragmented assembly offering infinitely more biological insight than almost any physical map.  Mapping would always be playing a supporting role, and that role was secure only as long as genomes remained hard to assemble.

So a huge blow to all the companies in this space was the breakout of long read sequencing for microbial genome assembly, led by PacBio and its collaborators.  With highly accurate assemblies of microbes becoming routine, a nice tester market for the mappers essentially evaporated.  Yes, a few really tough genomes still were out there and a few bugs appear recalcitrant to good long read preps, but mapping's role was now to deal with the scraps.  And as long read assembly has been shown to be computational feasible for large diploid genomes, the space left for mapping continues to shrink.  Rather than being enabling, mapping must try to be more cost effective, and as the long read methodologies improve in read length the number of spaces where mapping is attractive grow fewer and fewer.

One space that Nabsys was reaching for was oncogenomics, looking at large scale rearrangements in tumor samples.  If long read sequencing remains expensive on the scale of a human genome, then this would remain attractive -- if there was demand.  It hasn't been obvious there is a great demand here, based on the an apparent lack of high profile papers using these approaches on a large scale.  

What is truly threatening to all of the mapping companies is the rapid development of Oxford Nanopore's platform.  While the technology isn't an actual substitute for mapping at this time, reports of 100Kb+ reads should give any investor pause. Oxford's PromethION is about to hit the real world, and should it work then the best mapper is a sequencer.  I'm not saying here it will work (though I'll bet it will), but it represents a very real threat to the market.

Even if PromethION should do a face plant, numerous other long-range sequencing technologies have been hitting the market, as covered in this space.  Rather than being the only game in town to solve these problems, mapping becomes one of the pack -- still able perhaps to solve some problems unsolvable by the others, but a limited number of them. 

This is the unfortunate and brutal reality of the technology space: you can have a mind-blowingly amazing technology and simply be plowed over by one that is just a bit more so or by technologies that aren't quite as stunning but which fit into established practices.  Nabsys had a great idea to pursue, but the landscape shifted before they could get that idea into a fully launched product. Vowing to reboot the company is admirable from the tenaciousness it displays, but no amount of elbow grease and pluck is going to change the market realities.


Anonymous said...

Hi Keith,

Your statement: "Nabsys had developed an all-electronic single molecule physical mapping approach, though unfortunately many news reports mistaken referred to this as DNA sequencing.. " is not totally accurate. Nabsys was very much started as a DNA sequencing company. There are official quotes from company executives/directors saying as much and the old Nabsys logo even had the words "Electronic. DNA. Sequencing" as part of the logo. The notion was to use a large multitude of hybridization probes to sequence by hybridization via nanopores. I think that undertaking became more difficult than they were expecting (isn't it always?) and they pivoted to mapping as a less complicated version of the same core concept. Unfortunately, like you have said, the niche for mapping is shrinking with long-read technologies (true or synthetic/assembled) continuing to come onto the scene.

Keith Robison said...

That's a good point I should have made -- Nabsys in some sense saw sequencing as the ultimate in mapping and once the system was running would push the inter-probe distances smaller & smaller.

Unknown said...

Hi Keith,

As the "disgruntled founder" I was truly saddened to see the closing of NABsys. I disagree with your assessment on why NABsys failed, and on the prospects of competing technologies, however the place to address these scientific issues is in peer-reviewed scientific journals, not in sensational blogs.


Sean Ling

Keith Robison said...

Dr Ling,

I'm sorry that you view this space as sensationalist, as that is not my goal. But I agree with you that a good place for scientific information would be in the peer reviewed literature, where unfortunately there is a paucity of data on Nabsys' technology or its utility. And like it or not, companies must not only do interesting science but survive in the marketplace, and good science is not always sufficient to ensure survival.

AKatawazi said...


Thank you for providing great information, I regularly check in on your blog to see what is happening in the industry. I feel bad for Dr. Ling and his company but at the same time the rate of technological advancement makes it possible for new companies to get into the space with much lower startup costs thanks to the advances we are seeing. I hope you continue to write irregardless of the occasional "disgruntled" detractor.