Personally, I would've gone with "No juice? Then how will we ever get the vehicle moving fast enough for the road to turn on? This is what we get for driving the 35 mph speed limit."
To be honest, you're right; it's extreme given the current climate and how many enzymes are patent-protected. But I do want to make a point here: where should the line be drawn? How much 3D shape does code need before it becomes a blueprint? It seems to me, honestly, that this sort of thing should be protected by copyright, not patent law.
Right now people are getting a lot of low-quality patents for enzymes that would make no sense applied to physical devices; Clontech has one or more patent(s) on reverse transciptase for disabling part of the native enzyme. Knowing to disable that part requires a fair amount of expert knowledge and research, but actually carrying out the work can be done in a number of ways, and their patent lists specific amino acids to knock out. That means someone else can still get a patent for the exact same functionality if they happen to break the enzyme at a different critical point. Because the patents involve sequence-level jargon, they're trivializing themselves instead of focusing purely on functionality, which is why (a) Myriad was allowed to get exclusive ownership of a naturally-occurring (if abnormal) sequence, and (b) there are a lot of competitors in the enzyme market for things like DNA replication.
That competition may be good, but it only exists because the patents are too specific in a lot of cases. It exists in circumvention of the spirit of the patent system, not because the products themselves are really all that patent-worthy. Specifically for domain-disabling, in a lot of cases the critical residues to mutate can be determined systematically with little creative work... this is far, far away from the kind of ingenuity involved in designing a new rubber sole, car engine, or a sophisticated algorithm patent.
In my opinion, the bar for enzyme patents needs to be raised to the level of "novel mechanism of action," like a really fancy chemical reaction patent. If you're just knocking the out teeth of, rearranging, and truncating parts of nature, then what you're really doing is creating a remix of something pre-existing, like some kind of bizarre chemical found poem or (gods forbid) an AMV. It's essentially art, not invention.
The method can be described as "put the enzyme in conditions where it can operate naturally." The conditions are defined by the enzyme (i.e. nature), and for most enzymes there's no cleverness required to figure them out; just a bit of trial and error to determine what components of the original environment are important, so even if the natural defence is ignored, the method for getting reverse transcriptase to work would be unpatentable because it's obvious. Patents involving enzymes usually include either (a) a really heinously tricky replacement for natural conditions, such as developing a new molecule or mechanism to emulate an environment that's hard to recreate on a bench, or (b) an engineered enzyme. There are actually engineered reverse transcriptases which are patent-protected, but Myriad isn't an enzyme company; they just do testing. Their entire business model is dependent on exclusive control of information.
Also, even more obnoxiously, even if cDNA didn't exist naturally, you can produce an identical piece of DNA—atom for atom—using a copying and splicing technique known as PCR. It's slightly more inconvenient to do this, but people will do it anyway, and Myriad's shareholders seem to think they can sue them too.
I'm kinda using a more flexible definition of model organism here; if a sea slug provides a lot of information about some particular system then it should be called a model too. I really wanted to say that sequencing for diversity's sake, like ColdWetDog was implying, gives us very diminishing returns. When it comes to drug design, the amount of information that can be garnered from analysing large chunks of the tree of life falls off fairly quickly, despite how pretty the heat maps may look. A couple dozen nearby genomes that say (for example) a given residue is essential is not generally going to be greatly improved by doing the same comparison with a thousand—especially when the whole thing's going to be tested biochemically anyway. (Despite how kirovs seems pretty upset with me for saying this.)
In this case, the photocopiers can be found in almost every animal and plant, and a good deal of viruses. There's not much human innovation; it's just a matter of knowing what to make a copy of and how to use the photocopier.
I was trying to come up with one earlier, but I could only think of a car manufacturing analogy. We'll try it anyway:
There's this really good car design that's in the public domain, and it's really popular. Unfortunately, some of the factories that make the car have screwed up molds in some of the ducting that cause the airbags to fill up with motor oil at random. (Sophisticated, I know. Bear with me.) The loss of motor oil eventually causes the engine to overheat and the pistons to fuse, destroying the car. (Okay, so it's not a really good car design. It's a Lada. But still.)
To fix the car, you have to remove the airbags, but this is an obvious safety hazard and wildly undesirable, so no one wants to remove them if it's unnecessary. Determining whether or not they need to be removed depends on knowing the shape of the gasket that causes the leak: a tiny, hard-to-reach component buried deep in the gearbox that can't be checked without disassembling the car.
Aha, says Myriad, but we have a solution—our experts took a mold of the gasket, and if the mold fits on the gasket, you'll be able to tell if your car's at risk! You just stick the mold on the end of a crowbar and put it in a maintenance port that's normally too small for someone to reach inside. If the mold clicks on, then you know it's a match; otherwise, you're fine.
Brilliant! says every mechanic in the world, and they go out to make the mold, because a lot of them have seen these terrible cars and they have examples of the malformed gasket on hand.
But no! says Myriad, who has patented the mold. This is our invention, and you can't use it! Here is our patent.
But it's just a copy of a broken version of something in the public domain! The gasket's shape is already in our cars! says everyone else. In fact, there are molds just like the one you made in the factory! So they take Myriad to court, and the judge says:
1. Since the original mold that made the screwed up gaskets is in the public domain, you can't patent it.
2. But Myriad did work by making the mold and they're very clever, so I want them to be able to patent their invention. (Their lawyers convinced me of this.)
3. What really matters is the shape of the mold, though, not the mold itself. Myriad may not own the factory molds, but they own the shape. If you can cut up one of the factory molds and put it on your crowbars you're fine, and you can make copies of it too, but if you make your own the same way Myriad did, you're infringing on their patent.
4. In fact, we're just going to write this judgement in such an ambiguous way that you're infringing even if you use the original mold, because a copy of it is physically indistinguishable from Myriad's molds.
Remember the music videos at the end of Bill Nye episodes? That Bio-Rad commercial is like that for adults. Unfortunately, none of my professors sang along when they brought it up in class.
Most of the outliers used as benchmarks that I've seen are frogs, like Xenopus tropicalis. These genomes are generally sufficiently well curated and distant from our own to support arguments about conservation. It's excessive to worry about taking a census from the entire tree of life, particularly when the actual biochemists in charge of drug production don't trust bioinformatics anyway, and ultimately will carry out detailed structural analyses and mutation assays on the final candidate targets.
A very coarse estimate is sufficient, as evidenced by the huge wealth of drug targets that have been identified even without the consideration of a comprehensive survey of the tree of life. Certainly there is room for refinement, and perhaps you'll see a small improvement in predictive accuracy, but proving the point that the nice big protein sequence alignment figure in the errata of the paper has a bunch of Ws where we have a Y really only requires a dozen or so pertinent examples; the returns are very diminishing. If you have to sequence ten thousand obscure mammals just to find one that has a deletion mutation right where you want it, it's probably not representative.
While I'm grateful for your comments, I'm afraid you're late to the party; I'd advise you to read some of the subsequent comments. My point was that sequencing biological diversity was not pertinent to medicine. I intended to exclude model systems from my declaration, but that wasn't perfectly clear.
Yeah, it's one of the easiest things to kit since it's just an extra step on top of PCR. All of the big vendors supply it.
Given that the patented sequence is literally a PCR assay for a handful of single-nucleotide mutations, it would be technically impossible to not use the sequence in question, so they can be certain that such a case would go forward. From the patent:
All four sequence variants are heterozygous and each appears in only one of the kindreds. Kinderd 2082 contains a nonsense mutation in exon 11 (FIG. 9A), Kindred 1910 contains a single nucleoting insertion in exon 20 (FIG. 9B), and Kindred 2099 contains a missense mutation in exon 21, resulting in a Met.fwdarw.Arg substitution. The frameshift and nonsense mutations are likely disruptive to the function of the BRCA1 product.
Sorry, that was a bit of confusing language; I meant it to be read as "cDNA is a copy of the transcript."
And I think I have a better way of doing the metaphor than ROT13: different codepages. (Although the idea of ROT13 is apt to the physical reality, it's not really that important to the explanation.) Unicode genome, 8-bit transcripts, 7-bit ASCII proteins. Intergenic DNA is in Russian, transcribed RNA is in Greek, and RNA polymerase converts things to a Greek codepage.
Yes, it's relatively easy to PCR up the sequence from the genome once you know what you're looking for.
Let me make an extended computer metaphor to explain this properly.
The human genome is comprised of many different types of information, which we can liken to pieces of text in different languages. Consider a case where we have a long document written in Russian that discusses source code for a program that was heavily commented—but in Greek. (For those of you following along in the lab, Russian is the untranscribed DNA, Greek is transcribed but noncoding, and the ASCII program itself is the coding sequence.) All of this is in UTF-32.
When the cell functions normally, it uses the program's source as a build script: it copies the code out of the document into CP-1253 (the Windows Greek code page), and then cuts out any remaining "??????" left by the transcription, as well as all of the Greek comments in the middle of the code. There are some comments at the start of the code and at the end, but most of the copy is the code itself.
In the Myriad case, they've found a program that works the other way, called reverse transcriptase. It turns the 8-bit Greek/ASCII mixture back into UTF-32, but the Russian and the intragenic comments are still gone, leaving us with a usable sequence that we can do a bitwise search comparison with. This is very convenient to do because there's already a complete set of such sequences available. These are called cDNA libraries, and we rely on them everyday to tell us where the Greek comments are.
What researchers will be forced to instead is to cut the sequences out by hand directly from the wide Unicode source, using a lab technique called PCR splicing. The stupid thing is that there's no chemical difference between the two—it would be impossible to prove one way or the other where the diagnostic sequence came from, unless you knew the sequence's history. It just means that researchers now have to fumble around with splicing a bit more. Presumably the only reason Myriad isn't crapping their pants is because they think they can litigate based on intent in such cases.
The core enzyme in question, reverse transcriptase, occurs naturally in humans, and was first isolated in 1970 by two independent teams. The process is something like "put it in hot water with raw materials, your template sequence, and some salt." Even if a patent were awarded for it, it would have expired in the late nineties.
The only time cDNA occurs naturally is in the reproduction of a retrovirus or the replication of a retrotransposon. Myriad is still using the natural human sequence in their tests. This is like getting ownership of a quote from a book because you copied it into a Notepad window and then into Google, instead of just pasting it directly into Google in the first place, and then claiming it's a good way to find a certain rare edition of the e-book because it contains a typo made by the original author. It's completely and utterly intellectually dishonest.
It's disastrous. cDNA is just a direct copy of the most important part of what's in the genome—the actual transcript that gets used to make the final protein. This isn't a victory at all.
Given that a lot of the products of this work involve expensive drugs which both save lives and break banks, it's impossible to say for certain, although as I have a lot of colleagues in that pool I can say with some authority that they certainly think it's productive.
Slashdot Mirror
There was a story about how much Google resents the NSA six days ago, so I don't think you've hit the nail on the head here.
Licensees, but it's a labour of love, so, uh... something something. They asked for it. That's the important thing.
Personally, I would've gone with "No juice? Then how will we ever get the vehicle moving fast enough for the road to turn on? This is what we get for driving the 35 mph speed limit."
To be honest, you're right; it's extreme given the current climate and how many enzymes are patent-protected. But I do want to make a point here: where should the line be drawn? How much 3D shape does code need before it becomes a blueprint? It seems to me, honestly, that this sort of thing should be protected by copyright, not patent law.
Right now people are getting a lot of low-quality patents for enzymes that would make no sense applied to physical devices; Clontech has one or more patent(s) on reverse transciptase for disabling part of the native enzyme. Knowing to disable that part requires a fair amount of expert knowledge and research, but actually carrying out the work can be done in a number of ways, and their patent lists specific amino acids to knock out. That means someone else can still get a patent for the exact same functionality if they happen to break the enzyme at a different critical point. Because the patents involve sequence-level jargon, they're trivializing themselves instead of focusing purely on functionality, which is why (a) Myriad was allowed to get exclusive ownership of a naturally-occurring (if abnormal) sequence, and (b) there are a lot of competitors in the enzyme market for things like DNA replication.
That competition may be good, but it only exists because the patents are too specific in a lot of cases. It exists in circumvention of the spirit of the patent system, not because the products themselves are really all that patent-worthy. Specifically for domain-disabling, in a lot of cases the critical residues to mutate can be determined systematically with little creative work... this is far, far away from the kind of ingenuity involved in designing a new rubber sole, car engine, or a sophisticated algorithm patent.
In my opinion, the bar for enzyme patents needs to be raised to the level of "novel mechanism of action," like a really fancy chemical reaction patent. If you're just knocking the out teeth of, rearranging, and truncating parts of nature, then what you're really doing is creating a remix of something pre-existing, like some kind of bizarre chemical found poem or (gods forbid) an AMV. It's essentially art, not invention.
Pretty much. Say goodbye to HIV.
The method can be described as "put the enzyme in conditions where it can operate naturally." The conditions are defined by the enzyme (i.e. nature), and for most enzymes there's no cleverness required to figure them out; just a bit of trial and error to determine what components of the original environment are important, so even if the natural defence is ignored, the method for getting reverse transcriptase to work would be unpatentable because it's obvious. Patents involving enzymes usually include either (a) a really heinously tricky replacement for natural conditions, such as developing a new molecule or mechanism to emulate an environment that's hard to recreate on a bench, or (b) an engineered enzyme. There are actually engineered reverse transcriptases which are patent-protected, but Myriad isn't an enzyme company; they just do testing. Their entire business model is dependent on exclusive control of information.
Also, even more obnoxiously, even if cDNA didn't exist naturally, you can produce an identical piece of DNA—atom for atom—using a copying and splicing technique known as PCR. It's slightly more inconvenient to do this, but people will do it anyway, and Myriad's shareholders seem to think they can sue them too.
I'm kinda using a more flexible definition of model organism here; if a sea slug provides a lot of information about some particular system then it should be called a model too. I really wanted to say that sequencing for diversity's sake, like ColdWetDog was implying, gives us very diminishing returns. When it comes to drug design, the amount of information that can be garnered from analysing large chunks of the tree of life falls off fairly quickly, despite how pretty the heat maps may look. A couple dozen nearby genomes that say (for example) a given residue is essential is not generally going to be greatly improved by doing the same comparison with a thousand—especially when the whole thing's going to be tested biochemically anyway. (Despite how kirovs seems pretty upset with me for saying this.)
In this case, the photocopiers can be found in almost every animal and plant, and a good deal of viruses. There's not much human innovation; it's just a matter of knowing what to make a copy of and how to use the photocopier.
...sorry, for all transcribed DNA, not just coding DNA.
They're already effectively patented. This ruling changes little or nothing for coding regions of DNA.
But what if they then circumvent those laws? Maybe an anti-anti-circumvention law circumvention law is in order...
Maybe this car analogy will work better? :)
I was trying to come up with one earlier, but I could only think of a car manufacturing analogy. We'll try it anyway:
There's this really good car design that's in the public domain, and it's really popular. Unfortunately, some of the factories that make the car have screwed up molds in some of the ducting that cause the airbags to fill up with motor oil at random. (Sophisticated, I know. Bear with me.) The loss of motor oil eventually causes the engine to overheat and the pistons to fuse, destroying the car. (Okay, so it's not a really good car design. It's a Lada. But still.)
To fix the car, you have to remove the airbags, but this is an obvious safety hazard and wildly undesirable, so no one wants to remove them if it's unnecessary. Determining whether or not they need to be removed depends on knowing the shape of the gasket that causes the leak: a tiny, hard-to-reach component buried deep in the gearbox that can't be checked without disassembling the car.
Aha, says Myriad, but we have a solution—our experts took a mold of the gasket, and if the mold fits on the gasket, you'll be able to tell if your car's at risk! You just stick the mold on the end of a crowbar and put it in a maintenance port that's normally too small for someone to reach inside. If the mold clicks on, then you know it's a match; otherwise, you're fine.
Brilliant! says every mechanic in the world, and they go out to make the mold, because a lot of them have seen these terrible cars and they have examples of the malformed gasket on hand.
But no! says Myriad, who has patented the mold. This is our invention, and you can't use it! Here is our patent.
But it's just a copy of a broken version of something in the public domain! The gasket's shape is already in our cars! says everyone else. In fact, there are molds just like the one you made in the factory! So they take Myriad to court, and the judge says:
1. Since the original mold that made the screwed up gaskets is in the public domain, you can't patent it.
2. But Myriad did work by making the mold and they're very clever, so I want them to be able to patent their invention. (Their lawyers convinced me of this.)
3. What really matters is the shape of the mold, though, not the mold itself. Myriad may not own the factory molds, but they own the shape. If you can cut up one of the factory molds and put it on your crowbars you're fine, and you can make copies of it too, but if you make your own the same way Myriad did, you're infringing on their patent.
4. In fact, we're just going to write this judgement in such an ambiguous way that you're infringing even if you use the original mold, because a copy of it is physically indistinguishable from Myriad's molds.
And that's how fucked up the patent system is.
Remember the music videos at the end of Bill Nye episodes? That Bio-Rad commercial is like that for adults. Unfortunately, none of my professors sang along when they brought it up in class.
If it helps, there's a sequel, although it's not as generic.
Most of the outliers used as benchmarks that I've seen are frogs, like Xenopus tropicalis. These genomes are generally sufficiently well curated and distant from our own to support arguments about conservation. It's excessive to worry about taking a census from the entire tree of life, particularly when the actual biochemists in charge of drug production don't trust bioinformatics anyway, and ultimately will carry out detailed structural analyses and mutation assays on the final candidate targets.
A very coarse estimate is sufficient, as evidenced by the huge wealth of drug targets that have been identified even without the consideration of a comprehensive survey of the tree of life. Certainly there is room for refinement, and perhaps you'll see a small improvement in predictive accuracy, but proving the point that the nice big protein sequence alignment figure in the errata of the paper has a bunch of Ws where we have a Y really only requires a dozen or so pertinent examples; the returns are very diminishing. If you have to sequence ten thousand obscure mammals just to find one that has a deletion mutation right where you want it, it's probably not representative.
Which is even more horrific, because that's literally a software patent.
While I'm grateful for your comments, I'm afraid you're late to the party; I'd advise you to read some of the subsequent comments. My point was that sequencing biological diversity was not pertinent to medicine. I intended to exclude model systems from my declaration, but that wasn't perfectly clear.
Yeah, it's one of the easiest things to kit since it's just an extra step on top of PCR. All of the big vendors supply it.
Given that the patented sequence is literally a PCR assay for a handful of single-nucleotide mutations, it would be technically impossible to not use the sequence in question, so they can be certain that such a case would go forward. From the patent:
All four sequence variants are heterozygous and each appears in only one of the kindreds. Kinderd 2082 contains a nonsense mutation in exon 11 (FIG. 9A), Kindred 1910 contains a single nucleoting insertion in exon 20 (FIG. 9B), and Kindred 2099 contains a missense mutation in exon 21, resulting in a Met.fwdarw.Arg substitution. The frameshift and nonsense mutations are likely disruptive to the function of the BRCA1 product.
Sorry, that was a bit of confusing language; I meant it to be read as "cDNA is a copy of the transcript."
And I think I have a better way of doing the metaphor than ROT13: different codepages. (Although the idea of ROT13 is apt to the physical reality, it's not really that important to the explanation.) Unicode genome, 8-bit transcripts, 7-bit ASCII proteins. Intergenic DNA is in Russian, transcribed RNA is in Greek, and RNA polymerase converts things to a Greek codepage.
...the art of the car analogy is definitely dead.
Yes, it's relatively easy to PCR up the sequence from the genome once you know what you're looking for.
Let me make an extended computer metaphor to explain this properly.
The human genome is comprised of many different types of information, which we can liken to pieces of text in different languages. Consider a case where we have a long document written in Russian that discusses source code for a program that was heavily commented—but in Greek. (For those of you following along in the lab, Russian is the untranscribed DNA, Greek is transcribed but noncoding, and the ASCII program itself is the coding sequence.) All of this is in UTF-32.
When the cell functions normally, it uses the program's source as a build script: it copies the code out of the document into CP-1253 (the Windows Greek code page), and then cuts out any remaining "??????" left by the transcription, as well as all of the Greek comments in the middle of the code. There are some comments at the start of the code and at the end, but most of the copy is the code itself.
In the Myriad case, they've found a program that works the other way, called reverse transcriptase. It turns the 8-bit Greek/ASCII mixture back into UTF-32, but the Russian and the intragenic comments are still gone, leaving us with a usable sequence that we can do a bitwise search comparison with. This is very convenient to do because there's already a complete set of such sequences available. These are called cDNA libraries, and we rely on them everyday to tell us where the Greek comments are.
What researchers will be forced to instead is to cut the sequences out by hand directly from the wide Unicode source, using a lab technique called PCR splicing. The stupid thing is that there's no chemical difference between the two—it would be impossible to prove one way or the other where the diagnostic sequence came from, unless you knew the sequence's history. It just means that researchers now have to fumble around with splicing a bit more. Presumably the only reason Myriad isn't crapping their pants is because they think they can litigate based on intent in such cases.
The core enzyme in question, reverse transcriptase, occurs naturally in humans, and was first isolated in 1970 by two independent teams. The process is something like "put it in hot water with raw materials, your template sequence, and some salt." Even if a patent were awarded for it, it would have expired in the late nineties.
The only time cDNA occurs naturally is in the reproduction of a retrovirus or the replication of a retrotransposon. Myriad is still using the natural human sequence in their tests. This is like getting ownership of a quote from a book because you copied it into a Notepad window and then into Google, instead of just pasting it directly into Google in the first place, and then claiming it's a good way to find a certain rare edition of the e-book because it contains a typo made by the original author. It's completely and utterly intellectually dishonest.
I'm sure you think lots of things that make you special and unique.
It's disastrous. cDNA is just a direct copy of the most important part of what's in the genome—the actual transcript that gets used to make the final protein. This isn't a victory at all.
Given that a lot of the products of this work involve expensive drugs which both save lives and break banks, it's impossible to say for certain, although as I have a lot of colleagues in that pool I can say with some authority that they certainly think it's productive.