Tuesday, December 31, 2013

Peering Through the Flowcell Glass, Darkly

As 2013 draws to a close, I've decided to stick my neck out and make some predictions for 2014.  Perhaps I'll get lucky and a few will even come true!  After several mental experimentations on the structure, I'll settle for stepping roughly past each major player.
Almost a non-prediction is that Illumina will remain the 800 kilogram gorilla of the sequencing world, as there are no realistic threats to their top position in sight at the moment.  That means that Illumina's biggest threat is to trip over their own feet.  There's already been some grumbling over at SEQAnswers around the confused state of the different HiSeq models and which ones will be upgradable to 1Tb next year.  For the MiSeq, can Illumina continue to push fragment and read lengths to a kilobase or more?

Ion Torrent has bumped up against the reality that no matter how mature their semiconductor technology, the real challenge is in chemistry and biochemistry.  The new Hi-Q enzymes promise to help on the raw indel error side, but the critical Hi-Q for Proton doesn't yet seem to be unveiled.  The Proton's chip upgrades seem to be perpetually stalled in advancing the throughput, presumably because shrinking the wells and packing them close together is hurting signal-to-noise on two fronts: less signal and more bleed-over from adjacent wells.  Ion needs to also look at their acquisition strategy (presuming they have one); Illumina outwitted them first by snapping up the Nextera technology via Epicentre and then by snagging Moleculo.  Either one could have represented a possible counter-punch to Illumina and given them unique capabilities.  Perhaps Ion, if allowed by new parent Thermo, could go find a Moleculo-like company to gobble.

Pacific Biosciences had a spectacular 2013, becoming the gold standard for microbial genome sequencing and demonstrating a number of promising applications for other genomes.  PacBio continues to demonstrate the requirement of coupling of radically new sequencing approaches on important new algorithms; the HGAP assembler and Quiver assembly polisher were critical to their success, as were astonishing gains in read length and throughput.  A key challenge for 2014 is to find ways to squeeze more throughput.  With 20Kb+ reads starting become regular occurrences, the 50Kb-ish limit for DNA in solution (or else it snaps routine motion) starts to loom. Sample prep will continue to be a big focus for PacBio, especially as for microbes the sample prep cost outweighs the sequencing cost.  Also key will be squeezing out better implementations of their flagship algorithms (which are compute horsepower hungry), proven extension of those algorithms to diploid and polyploid genomes, and getting their software suite painlessly installable, even by ham-fisted bloggers.  But perhaps their biggest worry is that their hard work of blazing a trail for long, reduced accuracy reads, they've enabled dangerous competition.

The earliest such potential competition to PacBio is Oxford Nanopore, which had breathless coverage followed by a long silence.  Oxford has announced a beta test program like none before, open to the masses instead of a few handpicked genome centers.  It's a bold move, but bold can quickly be relabeled reckless should it hit snags -- and like none before means that surprises await. Still, this ensures that their novel technology gets out to those most likely to try something new.  Will the MinION work?  Personally I'm bullish, both because I'm naive enough to think they wouldn't announce such a program unless they were really ready to go (and after all, I did run a device to do something back in October) but also because I believe that the unusual pricing structure sets a very low floor for performance.  In other words, MinION may fit Clayton Christensen's definition of a disruptive technology, potentially underperforming in all current markets but opening up brand new markets that were previously blocked by cost.

2014 should be QIAGEN's coming out year for their sequencing solution.  The big Q's been very Quiet about Quantitative performance, releasing very little with regard to throughput or error rates.  Perhaps this year they, or some early customers, will Quack a bit.  Who might Quake?  Ion probably has the most to lose of the existing players, as QIAGEN is aiming straight for the clinical amplicon market that is still up for grabs.

GnuBio is another company aiming for the amplicon market, and like QIAGEN is promising end-to-end automation and user-friendly, results-oriented informatics.  But no Gnu has been seen outside carefully controlled environments; when will it be ready to roam the marketplace's plains?  That will decide whether 2014 will represent a Happy Gnu Year. 

2014 might also bring on other new players, such as GenapSys and Genia.  Will it?  Time will tell.

All these changes in the sequencing marketplace will have ripples elsewhere, particularly for the trio of companies that have built physical mapping platforms, OpGen, BioNano Genomics and Nabsys. Their challenge is to stay relevant in world where sequencing becomes faster, longer and cheaper.  One company offered me a "special" of mapping a microbe for $500; I can sequence an E.coli-class genome on PacBio for $800.  Mapping certainly hasn't gathered much brain share in the academic informatics community, perhaps due to a lack of standard data formats.  Mapping may still prove to be valuable for really hairy, high-ploidy plant genomes, but without good integration to sequencing methods few will opt to gen-erate maps.

About time to hit "Publish" and start thinking of what I left out of the above!


3 comments:

Anonymous said...

My prediction is that ONT is incapable of robust sequencing at this time, they have been spurned by the major genomic centers and ate going the Google Glass - we made it, we don't know what it is good for, so let's see if anyone can figure out how to do something useful with it. Watch out for ONTholes. The only wildcard is that they secured additional funding recently. Either they completely pulled the over the head of the investors with another stunt a LA AGBT or I am full of SHT. Given my tank record at prognostication both are equally probable :-)

Nick Loman said...

"they have been spurned by the major genomic centers" - evidence for statement?

Anonymous said...

I do not have hard proof, but if their tech really worked why wouldn't Broad go apesh*t over it and be the first to publish all the cool science people claim can be done with such. a tech. There really is no other explanation I can come up with. Can you?

And let's not kid ourselves with the democratization BS. ONT has sucked up so much capital already, investors would probably be booking gains if they could. Yes, this is all unsubstantiated speculation, but ONT 's official story can't seem to get past my BS detector.

And apologies for the mangled messages. Typing this on am phone keyboard is a PITA