Wednesday, April 24, 2013

What Might Knock Illumina Off Its Perch?

The big sequencing platform news this week is Roche's announcement that they are ending their collaborations aimed at developing new sequencing technologies.  A collaboration with DNA Electronics aimed to create an Ion Torrent-like system whereas a more ambitious effort with IBM aimed to create instruments based on "DNA transistors".  As noted by Forbes' Matthew Herper in a thought-provoking piece, Illumina is sitting very high atop the sequencer market.

Illumina has dominated sequencing for a number of years now, despite a number of challenges.  Having delivered a very good chemistry, particularly with the simple bridge PCR template preparation (as opposed to the emulsion PCR which has been the theme for all other clonal sequencing systems) and continued to improve on it. When Solexa was preparing the chemistry prior to being acquired by Illumina, the utility of 25 bp read lengths was explored; MiSeq now routinely does 2x250 paired ends and Illumina has discussed reads in excess of 400 bases.  Illumina blunted the challenge of Ion Torrent by delivering the MiSeq, and continues to be the preferred platform from a quality point-of-view.  Masses of informatics either specific to Illumina or tuned to the platform dominate the open source arena.  Some of this is a bit of luck: because Illumina's chemistry, as opposed to most once or current competing chemistries (454, Helicos, Ion and PacBio), has a very low insertion/deletion ("indel") error rate, large numbers of programs based on counting short words ("kmers") have proven highly valuable.  

Given Illumina's dominance, it is sometimes hard to imagine them being knocked into a trailing position, but the reality is no market leader dominates permanently.  Below I outline some possible scenarios for Illumina falling out.

Someone Else Launches A Better Technology 

The obvious scenario is the release of a new sequencing technology which beats Illumina for important markets.  One question Herper raised is whether sequencing is cheap enough now for the majority of applications.  Particularly in medical applications, other components of a diagnostic process (sample acquisition, result interpretation) are being viewed as generating more cost than the sequencing itself.  

It's hard to picture any of the current player seriously taking on Illumina, with the outside exception of a revitalized Ion Torrent.  But, even before the sale of Life Technologies was announced there was a perception that Ion's stunning development pace was slowing down.  Ion must also overcome the perception of lower quality data.  Roche's announcement of abandoning new technologies is seen by some as a harbinger of Roche leaving the sequencing business altogether; 454 technology has just not kept up and is being surpassed on every performance dimension.  If 454 is limited to an installed base constantly tempted by ever improving competing platforms, then its market will stall and then proceed to oblivion.  Pacific Biosciences is a wonderful platform for generating long reads, but the huge upfront cost and higher cost-per-base will keep them a niche player.  A proven platform but with significant market challenges, few would be surprised if Pacific Biosciences is purchased by a larger player.  SOLiD is fading from view, though certainly still used.  Perhaps if Life Tech hadn't acquired Ion Torrent then this platform would have developed further, but its potential advantage (higher accuracy) has never caught the imagination of the sequencing community -- or at least hasn't overcome its perceived flaws.  QIAGEN's instrument isn't quite here and looks too much like a me-too to really threaten Illumina, though it may certainly steal some sales. In addition to QIAGEN, GnuBio is also likely to nibble away at the clinical market, at least those clinical tests which use PCR targeting.  I believe whole genome and whole exome diagnostics are showing too much power to be displaced by PCR tests, relegating the latter to niche status.   

So it would take a radically new chemistry to give Illumina a hard shove.  Nanopores are the obvious candidate, but the world still awaits data to back up the promise.  IBM has declared a commitment to continue developing their technology, but they'll need a new partner with expertise on the life sciences side.  Even if a really stunning technology came as a bolt from the blue, time would be required to gain acceptance.  The market has clearly become jaded from various sequencing companies (e.g. Oxford Nanopore, Pacific Biosciences) making announcements far ahead of demonstrating day-in, day-out utility and performance.  And, there is always the possibility that it is Illumina that launches, or acquires, the next great sequencing technology.

Arrogance or Complacency

These are really two sides of the same coin, though perhaps expressed differently.  If Illumina becomes convinced that they cannot be challenged, then they are likely to make mistakes.  Annoying customers and partners by arrogance is one route; dropping vigilance is another.  Ion will continue to nip at the leader's heels; giving them an opening could enable substantial share to be lost.  Previous serious issues with Illumina's reagents should be a constant reminder that Illumina's dominance was never set in stone; failure to fix these issues could well have enabled SOLiD to carve out a bigger base.   Arrogance could take multiple forms.  High-handed treatment of customers.  Failure to aggressively develop the platform. 

Antitrust Activity

In a sense, Illumina needs its competitors, because otherwise they will be open to antitrust attacks.  Indeed, IIllumina's empire building is already raising some eyebrows.  In particular, their acquisition of Verinata puts them in direct competition with a number of large customers, and raises the spectre of a vertically-integrated genomics behemoth.  The worst case scenario here would be active prosecution of antitrust complaints by either the U.S. or E.U..  Many powerful companies, from Standard Oil long ago through IBM and AT&T in my childhood  and Microsoft recently, have had their business courses tremendously changed by antitrust activity.  Even the possibility of such action is a serious issue and management distraction.  Even if actual lawsuits did not emerge, the perception of an Illumina near-monopoly might drive a "stop ILMN" alliance of other players.  

Plain Bad Luck

The violent events of last week in the Boston area (the 7-11 holdup and murder of MIT police officer Sean Collier occurred just around the corner from my workspace; the gas station near two previous workplaces) underscore the difficult times we live in, though perhaps all human history is characterized by difficulty and unpredictable large events changing the history of companies and nations.  Some unfortunate event utterly outside Illumina's control could affect them greatly.  Apple has hardly gone under, but it would be difficult to find someone knowledgeable who would claim that a pancreatic neuroendocrine tumor didn't change the course of the company.  The terror attacks on 9/11 were not directed at biotech, but helped destroy the bioinformatics firm Lion (their CFO died in the attacks) and the resulting stock market plunge clobbered the ability of many companies to raise money.  I hope no such things affect Illumina, but the truth is that random events drive our lives far more than we like to admit.  I'm sure companies such as Illumina attempt to plan for such occurrences, but there is a limit to what can be anticipated.

Concluding Thoughts

How long will Illumina's reign atop the sequencing world last?  I'd say anyone's guess, though  it would seem that no major shift is likely for at least two years, just thinking that a new technology would need 1-2 years to gain acceptance and accumulate large market share.  Still, if a really radical technology arrived that was end-to-end (instrument, sample prep, data generation) cheap, had high quality and decent read lengths, I'd be elbowing to be first in line to replace my Illumina data with the new kid.


Unknown said...

Great post as always! I work in this industry as a software developer and it is vital to see the trends. We had a bet on SOLiD that did not work out. All that effort in vain.

Anonymous said...

My view is in a year or two, the only one who has the chance to dethrone Illumina is oxford nanopore (well, Illumina still holds some of its shares IIRC). Personally, I am quite impressed by the progress Illumina has made since 2008. Other technologies are not bad. Just they have been outpaced by Illumina. I have seen Illumina's 400bp data. The tail is not as good as 250bp, but probably Illumina will overcome the problem internally in several months.

Anonymous said...

I enjoyed the article, but it seems a bit misleading to compare current Illumina technologies to the Ion Torrent and not even mention the most current Ion platform (the Ion Proton). I'd be curious to know how much the market Ion Proton could consume if they continue on their current trajectory.

Keith Robison said...

WRT Ion, I conflated Torrent and Proton. The error profile is very similar & Proton has not stormed ahead in a manner to threaten Illumina on cost-per-base. If they really reach their lofty goals, then there might be some discussion, but for whole human genome sequencing Proton isn't yet economically competitive.

Anonymous said...

I'm surprised not seing any mention of nor two very short term MySeq killers

Keith Robison said...

Neither Genia nor Genapsys has shown any actual sequencing to my knowledge. Genia also changed their chemistry plan late last year.

Even if either of these (or Oxford Nanopore, or all the other exotic sequencing tech companies out there) could demonstrate actual function by the end of this year, it is another thing entirely to gain market acceptance & scale up your manufacturing to the scale that MiSeq is at, let alone HiSeq.

Brian Krueger said...

As always, spectacular post. Although, I think it's probably safe to mention that Illumina is probably safe for the foreseeable future. Any new tech is going to have to prove itself, ramp up production and work through all of those QC/QA issues. So it will likely be 2 years from the announcement of useable tech before any of us will switch.