Monday, January 02, 2017

Sequencing Technology Outlook, January 2017

Another year of blogging is upon us!  Since the J.P. Morgan Conference starts a week from today and then before long it's time for AGBT.  So if one is going to prognosticate, then there's no time to lose, as announcements could start flying at any time.
A gigantic wildcard for any 2017 predictions is the change in U.S. administration.  Any change in the government can produce great changes in policy which can ripple into science.  The incoming President-elect has been nothing but unpredictable, so I'm apt to disbelieve any precise prediction.  But with possible changes to FDA policy, corporate tax policy, immigration policy and trade policy, this could be a year full of surprises as companies respond to changes.  Since most political discussion yields far more heat than light, I (mostly) won't say more on this topic, but I feel it must be mentioned.


Illumina remains the market leader and is widely perceived as having a near monopoly, but lately has not had much luster in the markets.  A perception has been growing that the market for high-throughput sequencing may be saturated, as noted in blog posts by James Hadfield and Dan Kobolt.  At the low end, Illumina launched the MiniSeq last year but there has been little buzz about it in the research markets.  Supply chain and reagent quality issues have  cropped up, ranging from shortages of NextSeq high output kits to problems with data quality on the 2x300 kits for MiSeq.  Technical problems have also become apparent with the exclusion amplification chemistry, leading to incorrect barcode assignments.  There are also persistent rumors that the NeoPrep library preparation robot is not winning raves.   Illumina changed CEOs last year, possibly altering direction.  Finally, Illumina at times seems more focused on their new ventures Helix (personal genomics) and Grail (cancer liquid biopsies).

The one thing we definitely know to look for from Illumina is the new Firefly instrument, a non-optical take on sequencing-by-synthesis to yield a smaller, more portable box at a lower price.  The original announcement put launch of Firefly in the second half of 2017, though that could be met with a launch on the next New Year's Eve.  So Illumina is a bit under the gun to unveil the system at either J.P. Morgan or AGBT.  Will this just be a lot of publicity shots, or will working prototypes or even data be available?  

With regard to the perceived sequencing glut, Illumina could try to improve HiSeq X5 and X10 utilization by removing all restrictions on the types of samples which can loaded.  Realistically, libraries are libraries and these restrictions (which really mean Illumina won't replaced bad reagents if outlaw libraries are loaded) have always rankled their owners.  

It would also be interesting to see Illumina respond to the growing use of long read sequencing technologies.  The Moleculo technology really isn't a long read tech, restricted to under 10 kilobases (often more like 5 kilobases in practice).  Illumina might finally commercialize their own linked read technology, or could go out and buy one of the small linked read technology companies (iGenomiX, 10X Genomics or Dovetail).

Alternatively, Illumina could signal a further shift in attention away from research to downstream applications by launching yet another spinoff in the likeness of Grail and Helix.  

Pacific Biosciences

From a research perspective, PacBio is truly the number two company right now, becoming increasingly used to build high quality de novo assemblies of even very complex genomes.  They were rattled recently by Roche abruptly terminating their partnership to drive the Sequel instrument into the clinic.  PacBio faces growing competition from Oxford Nanopore in the long read market and must defend that flank while pivoting into their own strategy for clinical applications.  I do feel a need here to stray from my earlier promise to avoid discussing politics; it seems likely that changes in the FDA leadership are likely to lower the regulatory hurdles for launching laboratory-developed tests (LDT), the market PacBio is targeting.

So PacBio's requirements, from this perspective, are to demonstrate traction in the LDT space while continuing to show regular improvements in Sequel performance so as to avoid defections to Oxford Nanopore.  As will be discussed before, PacBio's random error profile remains their major advantage over ONT, so for applications such as de novo assembly PacBio retains a sizable edge.  

Oxford Nanopore

Oxford has been continuing to gain publications and launch new extensions of their technologies; expect to see a lot of ONT stories here in 2017 (including updates on hands-on experience, though not necessarily my hands).  Oxford's challenge, as always, is to keep delivering on their fantastical promises.  They've done this far more than some critics give them credit, but it will continue to require attention.  I've written recently on my skepticism of the PromethION instrument and the danger it will suck resources away from other efforts; 2017 should be a watershed year for PromethION and an opportunity to make those worries seem very silly.  Excited tweets of VolTRAX deliveries appeared last month; this year we'll start seeing user feedback on this sample prep instrument.  Similarly, the direct RNA sequencing kits started being seen in Twitter feeds at the end of 2017; success of these kits will be shown by the number of preprints generated by outside labs.  

ONT will also need to keep focusing on their commercial side, cutting delivery times and costs while maintaining a high record of customer satisfaction.  They've promised to start moving the MinION down the path of a diagnostics device; that will be one key step towards driving nanopores into the clinic.  But clinical uptake may also require guarantees that current chemistries will be available long-term, which would conflict with ONT's sometimes dizzying pace of chemistry development.  After all, 2016 saw three different DNA sequencing chemistries in user hands.  Oxford will presumably remain a target for intellectual property challenges, having fended off one from Illumina and now engaged in one brought on by PacBio.


Roche's Genia nanopore platform made several important advancements in the literature space in 2016, but the question remains when Roche will launch the system.  A successful Genia launch could be uncomfortable for just about all the existing players, as the publications suggest long reads with very low (and potentially random) error rates but with low reagent costs on a simple instrument.  Of course, until Genia actually launches and we see solid data as well as price lists, we won't know the actual specifications.  

An interesting new development is that University of California is suing Genia's founder Roger Chen, claiming misappropriation of intellectual property.  I plan to cover that in detail in another item ("plan" being a euphemism of being assigned the task by Mick Watson).  My expectation is that this suit will result in a change in the financials of Genia and Chen, but I'm skeptical that UC would try to sink Genia since seeing it succeed would be more profitable.  


It's a bit unfair of me to lump everyone else into one bin, but there's just limited excitement and/or visibility on the rest of the market.  Heck, SOLiD is technically still around and could be covered.  Ion Torrent and QIAGEN both appear to be focused on the targeted panel markets.  Ion's decision to abandon the difficult development of chips for whole human genome sequencing doesn't look so bad now that the market appears saturated.  QIAGEN is a sphinx, completely uninterested in the research market or in educating the world in the technical details of their instrument.  I had the pleasure of lunch recently with a principal at SeqLL, and they are happily focusing on niche markets where their simple sample prep and minuscule input requirements give them an edge.  I've heard that the ex-GnuBio technology at Bio-Rad is still alive, but until that is launched there's not much to talk about.  Similarly, GenapSys is rumored to still be working on their chemistry, but until that surfaces it is an enigma.  BGI and Complete Genomics are still out there, though BGI hasn't tried to sell their sequencers in the U.S. as far as I know, perhaps for IP reasons.  Then there are mapping companies such as BioNano Genomics and Nabsys 2.0, which remain complementary to sequencing technologies.  There are numerous other sequencing technology companies in varying levels of stealth and mystery; hopefully at least one will reveal themselves during the year to keep life interesting.  But until then, not much to say

I also expect 2017 to continue to have interesting developments in technologies at the periphery of sequencing.  I've mentioned linked reads above, which is a hot field, but other sample prep angles are out there.  In any case, I expect 2017 to be a big year in genomics, and look forward to trying to keep up with the field.


Kyle Serikawa said...

Thanks for the terrific snapshot Keith. Perceptive as always. I think one really key point about Oxford that you make is how diagnostic development and rapid platform evolution are incompatible. Even with the shorter timeframe for diagnostic development and clinical trial data collection, the Oxford practice of rolling out several updates a year to chemistry and platform basically make Oxford a non-starter even for an LDT.

It'll also be useful to see what happens with Illumina. Clearly making a bit on the lower margin but potentially more consistent revenue stream from sequencing based diagnostic tests will become a larger part of their biz. As with everything, comes back to biology: we don't know if early detection in the blood will help with cancer prevention/treatment--or if it'll just end up being an expensive test with no comparative effectiveness. Time will tell.

Anonymous said...

yes, lets all bash Oxford again.

of course you wouldn't use agile release methods on a clinical product you'd just freeze a version and make a separate branch of the technology and treat your research products+customers updates differently from the clinical ones, even organise your company around the specialization.

Keith Robison said...

Bashing? Just an obsevation of something ONT needs to do and also a mistake that many other sequencing tech companies have made in the past (I remember SOLiD burned a number of groups by prematurely scotching a chemistry). It also raises costs and complexity, keeping multiple chemistries in production. Not insurmountable hurdles, but nonetheless necessary for success.