Thursday, February 25, 2016

Digging into the Illumina Lawsuit vs. Oxford Nanopore

Illumina's and University of Washington's filing of a patent lawsuit and related trade complaint against Oxford Nanopore made big news yesterday, with nice coverage from Mick Watson, GenomeWeb, Nature's Erika Check Hayden, Technology Reviews' Antonio Regalado,  BioIT World's Aaron Krol, and venture capitalist Vishal Gulati. Each of these covers the onetime partnership between the two companies and their acrimonious parting of ways.   Oxford Nanopore released a short and pithy response. Having failed to get an early jump on things, the ground is already well plowed.  So my sloth and inertia have forced me to take an unpleasant route I usually spend great effort avoiding: actually reading the complaints and the two key patents licensed from Jens Gundlach's group at University of Washington (US8673550 and US9170230 ) they cite.

Now of course I am not a patent lawyer and have no training in that space beyond the informal admonitions of attorneys at my string of employers.  Furthermore, particularly in the U.S., patent law is complex and has been changing over the last few years (with the laudable goal of better aligning it with other legal systems, but the transition is confusing and painful).  Not my idea of an avocation -- nor do I go out of my way to engage in this in the workplace.  I still have nightmares from my Codon Devices days when I tried to figure out very precise synthetic constructs described only by a series of classic restriction digestion and ligation (sometimes with fill-ins or blunting) in a Russian doll set of patents.  And just a reminder: while I use both companies' technology and try to stay on friendly terms with each, I have no financial stake in this outcome nor any ties to either company (though via my employer I was a participant in MAP -- which perhaps I should stop stating given some of what I mention below).

Mick put it very well: nobody who does sequencing is going to like this.  Both Illumina and Oxford have exciting technologies that advance science, and each has its own strengths and weaknesses.  Indeed, the language in the trade complaint has galling language claiming that Illumina, PacBio and LifeTech machines could fill in for Oxford if MinIONs and PromethIONs were excluded from the U.S.; no other instrument has the portability of a MinION.  Illumina claims in the filings to be working on nanopore sequencing, but Oxford actually has working devices.  Plus, the filings make the claim that every individual who has operated a MinION device, even just for burn-in, has engaged in patent infringement.  While they are mostly using this to claim that ONT has serially lured others into misdeeds, this is definitely an idea which will not sit well with many scientists.

But this is business based on science, not science.  Patents are an important, if not always popular, part of that world.  The same patent law that is being pointed at Oxford is also one they count on to keep someone else from simply copying their technology.  Patent law is a complex and weird space, which is both why I generally stay out of it and view with great skepticism (often punctuated with snorts) when I see non-experts attempting to predict the outcomes of these battles.  Hopefully I won't induce many snorts with this piece.  Though if you want some chuckles, then some in the genomics crowd found bits of humor.

Before we get to the filings, one interesting bit has been the thought that Illumina launched the actions to foul up Oxford going public, a rumor which apparently broke recently (though I missed it). However, Oxford's CTO Clive Brown has denied that this was the case.

The patents which Illumina licensed from University of Washington cover the use of native and mutant MspA pores from Mycobacterium smegmatis, as well as homologs (more on that in a bit).  A key question is whether Oxford is using pores covered by these patents in their devices.  As Regalado uncovered, an early patent held by Oxford mentions MspA, but only very briefly.

The lawsuit filing cites several lines of evidence to suggest they are, some circumstantial and others more suggestive.  Oxford has never disclosed the identity of their pore(s), keeping them a trade secret, and has often claimed to have a large inventory of possible pores proteins.  I must admit to being a bit disappointed that Illumina's filing doesn't contain any references to flowcells left at dead drops or other exploits lifted from the pages of Alan Furst or John Le Carre.

The flimsier evidence is a number of comments made to the U of Washington inventors by Oxford or their academic advisors.  For example, when Gundlach published on MspA, Oxford inquired about licensing the patents and had nice things to say about Akeson's  Gundlach's work  Of course they did! It's bizarre to suggest that asking to rent something is evidence that you are planning to steal it, but that is the logic.

The filing also cites public comments from Oxford adviser Mark Akeson on a Gundlach paper, in which Akeson lauded the paper as shown below

Again, this seems to be very grey.  Akeson could have carelessly been tipping important information about the closely guarded nature of Oxford's pores, but alternatively he could have just been saying the work assuaged concerns that nanopore sequencing might never work. Which interpretation you take probably reflects much on your priors and biases.

Another line of circumstantial evidence shading into more serious questions can be found further down in the complaint

The apparent quote for Hagen Bayley would seem leave a lot of interpretive room -- "deduced in outline" certainly doesn't guarantee that existing patent filings provide a complete recipe for working out how MinION works.  On the other hand, the interpretation advanced by Illumina could be true as well.

The complaint also cites the fact that Oxford filed for an inter partes review, which Wikipedia explained to me is a formal process for challenging the validity of a patent.  Again, this could mean that Oxford wanted the patents dead so they could use MspA, but it could also mean they wanted them quashed because of the nebulous scope of a patent (I'll cover still later, but coming!).

Perhaps getting a bit stronger is Illumina's claim that many of ONT's recent patent filings use MspA nanopores in their working examples.  However, the extent to which one can build on another's patents is certainly not something I understand correctly, but patents are intended to teach their methods and be a basis for future innovation.  A sure sign someone is clueless on patents is when they say someone patented something to keep it secret; except in national security cases, patents require disclosure.  Also potentially troublesome item cited by the complaint is a recent Harvard pre-print which credits ONT with supplying mutant MspA protein used in the experiment.

Suppose we did know what pore Oxford was using and it is not precisely MspA or the specific mutants described in the U Washington patents, would it then be easy to determine the merits of the suit?  As Mick asked, it boils down to the question of what variation on MspA is covered by the patents, and that is a very, very grey area indeed.

One could imagine a patent world in which only the invention strictly described in the patent is protected, and any deviation from that description is allowed.  That could easily lead to absurd consequences, in which I do something trivial to your invention to claim my device is outside your patent.  This is covered, as I understand, by the doctrine of equivalents, which allows for variation.  However, what is allowed for variation depends on the particular country's patent law (this Wikipedia makes clear), and even then there is still great subjective interpretation by the legal system.

When filing a patent, the lawyers try to nail down a lot of that variation. For example, if I am patenting a new type of meter stick, for which rigidity is critical, my patent might both specify that the material is rigid and then enumerate a long list of possible materials from which my device might be made, many of which would in turn be fuzzy descriptions (an iron alloy containing no less than 10% and no more than 90% cobalt).  You might be tempted to try to claim the whole universe ("constructed of any combination of atoms"), but that bumps into other problems -- if your claims encompass things which cannot be patented, such as that which exists at the time of your invention ("prior art"), that can be grounds for voiding that claim in your patent.  Hence, as I understand it, patents are broken into lots of claims and subclaims, so that invalidation of one part doesn't nuke the entire patent.

This is evident in the U Washington patents, using a pretty much standard approach.  For example, the '550 patent has the following language,
 Specifically disclosed are polypeptides which have at least 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99 percent identity to a wild-type MspA and wild-type MspA paralogs or homologs (e.g., wild-type MspB, wild-type MspC, wild-type MspD, MppA, PorM1, Mmcs4296), and mutants provided herein.
The '230 patent has similar language with a much longer list (and table) of paralogs.

So if your pore has 70% identity to MspA or any of the other named proteins, it might be covered by the patent.  And yes, in case you were wondering, various specific methods to calculate these identities are provided.  Though maybe not -- it could be ruled you are trying to grab too much space (I think that the examiner could raise that objection, but now that the patent is issued they are fixed -- but if anyone who really knows this stuff could chime in that would be great).  But also if Oxford can prove that at the right time (filing? I really, really don't understand the different types of dates enough) that pores were known with 75% identity, they might be able to erase the 75% and lower cutoffs.  Whether just showing such pores were public knowledge (it doesn't matter if publication was in an obscure journal or even just a poster at an open scientific meeting) is sufficient or some functional information would have need to have been known then, well again we hit the limits imposed by my ignorance.

The Washington patents also delineate specific molecular architectures and dimensions.   In the '550 patent:
A “vestibule” refers to the cone-shaped portion of the interior of an Msp porin whose diameter generally decreases from one end to the other along a central axis, where the narrowest portion of the vestibule is connected to the constriction zone. A vestibule may also be referred to as a “goblet.” See FIG. 1 for an example of the vestibule of a wild-type MspA porin. The vestibule and the constriction zone together define the tunnel of an Msp porin.
This is used to set up another whole range of claims (with the added fuzzifier of "about" on many of the measurements -- is there a legal definition of "about" at the nanometer scale?)

A “constriction zone” refers to the narrowest portion of the tunnel of an Msp porin, in terms of diameter, that is connected to the vestibule. The constriction zone of a wild-type MspA porin is shown in FIG. 1 (labeled “inner constriction”). The length of the constriction zone may range from about 0.3 nm to about 2 nm. Optionally, the length is about, at most about, or at least about 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0, 1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 1.8, 1.9, 2, or 3 nm, or any range derivable therein. The diameter of the constriction zone may range from about 0.3 nm to about 2 nm. Optionally, the diameter is about, at most about, or at least about 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0, 1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 1.8, 1.9, 2, or 3 nm, or any range derivable therein.
 More sections explain about possible amino acid modifications or uses of atypical amino acids (including beta-amino acids), always with expansive language such as "The following table shows non-limiting examples of unnatural amino acids that are contemplated herein".  The patent also goes forth and makes freshman year definitions, such as "Non-limiting examples of side chains that may be positively charged include histidine, arginine, and lysine."   The '230 patent appears to cover conservative substitutions in the pore (one reason for all those textbook definitions); i.e. they appear to trying to cover attempts to rewrite the MspA porin in a functionally-neutral manner.

So what does this all mean for the sequencing world?  First, there is the question as to whether Illumina gets their trade action -- if so, then no more MinION or PromethION in the U.S.  I haven't a clue how likely or quickly this could happen, but it is a scary thought.

Undoubtedly, the case will be long and costly -- lawyers are pricey.  Cases can be lost by simply running out of money or nearly so; Codon simultaneously folded and threw in the towel on a patent lawsuit.  Whether they could have won the suit (Codon was the defendant) will be forever unknown -- and the SOB plaintiff company (Febit) dropped the line of business they were suing over immediately afterwards.  Yup, patent fights easily start resembling a military siege such as Stalingrad or Verdun in terms of becoming a battle of attrition, thankfully measured in currency not lives.  But that is always money that could be going to innovation but is instead going to the legal profession, plus two companies burning attention and time on the conflict.  While unlikely, the possibility of both companies suffering cannot be ignored: Polaroid successfully used patent litigation to drive Kodak out of the instant photography market, but the wounds from that battle helped send both companies towards oblivion.

That's one reason I could imagine a side-effect of this being Oxford inking a partnership with another big player in life sciences.  Oxford has tried to go solo ever since dropping Illumina, but partnerships mean sharing the wealth and potential for distractions and interference, but a big partner with big pockets and their own legal staff could be very useful to have should this patent battle drag out.  Partnering with a large company skilled in reagent kits could also improve Oxford's performance in delivering devices and consumables, which has been a source of muttered complaint in the MinION community.  But, obviously this in not a position of commanding strength from which to negotiate, making such a step even more distasteful than it probably already was. 

Some commentators have noted that Illumina has been successful with patent lawsuits in the past, but this track record (based on a quick Google search) is not perfect: they eventually settled with a company called Syntrix over array patents.  Settlements with patent cross-licensing is a common outcome for these suits, though often after protracted litigation.  The previous bad blood between the companies doesn't bode well for any quick resolution. And could Illumina's sharp actions and dominant market position lead to antitrust suits either here or in Europe?  Time will tell, and about the only thing I would bet on is that this will all get more confusing and more expensive before any resolution is achieved.


Anonymous said...

It's clear that the ISIS patent filed two years before the U. Wash patent cites MSpA's use for sequencing.

How is identity measured? Changing one amino acid eg exchanging a cysteine for a lysine, would change the architecture of a pore massively. So amino acid sequence would be in the range claimed (or would it, is 99.6% covered?) by the physical appearance would be below the range depending on how a numerical comparison can be made.

Of course this all depends on whether Oxford are using MSpA. They've been fully aware of FTO for some years.

Anonymous said...

"more likely than not", that's the basis of their complaint to the trade commission.

the purpose of the action could be simply to provide an argument to ban the importation of the devices, disupting supply, stalling and buying time until they can think what to do next. probably not because of the rumoured IPO or the minION but because of the promethion as a directly competitive device to their suite.

when a company is so dominant it can sell expensive instruments and then dictate what can be sequenced surely that is a problem for users.

Anonymous said...

I've only seen one other article that notices that ILMN had a board observer seat at ONT at the time they secretly licensed the patents in question...

Anonymous said...

"Oxford inquired about licensing the patents and had nice things to say about Akeson's Gundlach's work Of course they did! It's bizarre to suggest that asking to rent something is evidence that you are planning to steal it, but that is the logic"

that's not logic, that's stupidity.

simply explained as an defensive, conservative patent strategy covering all aspects of an IP area to give commercial security. perfectly reasonable and well established IP strategy. if a company is not doing this they either don't have the funds or they are incompetent.

Anonymous said...

Having been in the field, and knowing well the status of the protein pore research (there hasn't been another pore published or patented, capable of resolving 5-6 bases since mspA), here are two possibilities:
- Illumina will win the case, force Oxford to pay a ridiculous license fee, crush Oxfords valuation, and pick them up at a fire sale price.
- Oxford moves the courts to set new precedent over what constitutes 'identical' proteins, and what is patentable about proteins. The Washington and Alabama applications cover a lot of ground in capturing the msp pore,so this may be the less likely option.

Anonymous said...

"Illumina will win the case, force Oxford to pay a ridiculous license fee, crush Oxfords valuation, and pick them up at a fire sale price."

Possibly, but the said patents, if applicable, only have jurisdiction in the US. As most of Oxfords business seems to be non-us it seems hard to belive that royalty in the US would "crush" a valuation.

Anonymous said...

Here is an interesting posting on Oxford Nanopore:
Looks to me, judging by the inline Glassdoor comments, that even if they beat off Illumina, their management will ensure their downfall. Furthermore, the Glassdoor comments very much accord with what I've heard on the grapevine.

Anonymous said...

"Here is an interesting posting on Oxford Nanopore:"

thank you samanta. we ve all heard a few things about you too.

Anonymous said...

As a follow-up, it seems two claims have made it through the review:

leaving claims 17 and 18 which I think are from this patent:

which appears to be for an msp* pore in a bi-layer reading ionic current, and for any equivalent systems using an mspA mutant.