Gene Genie

October 19, 2009 10:53 AM

I.B.M. Joins Pursuit of $1,000 Personal GenomeThe target is remarkable given that the original Human Genome Project successfully sequenced the first genome less than ten years ago and cost roughly $500 million to $1 billion. Advances in sequencing technology puts Moore's Law to shame: "In the last four to five years, the cost of sequencing has been falling at a rate of tenfold annually, according to George M. Church, a Harvard geneticist. In a recent presentation in Los Angeles, Dr. Church said he expected the industry to stay on that curve, or some fraction of that improvement rate, for the foreseeable future."

The old way of doing things was laborious. The new way involves "a 'DNA transistor,' which it hopes will be capable of reading individual nucleotides in a single strand of DNA as it is pulled through an atomic-size hole known as a nanopore. A complete system would consist of two fluid reservoirs separated by a silicon membrane containing an array of up to a million nanopores, making it possible to sequence vast quantities of DNA at once...[the goal is to build a machine that could sequence] an individual genome of up to three billion bases, or nucleotides, 'in several hours.' "

posted by storybored (47 comments total)

A great example of how science is mostly both fascinating and scary.
posted by glaucon at 11:03 AM on October 19, 2009

I think there is a world market for maybe five genome sequencers.
posted by DU at 11:05 AM on October 19, 2009 [17 favorites]

Oh yes, a quick animated description of chromosomes, DNA and how it all fits together.
posted by storybored at 11:06 AM on October 19, 2009

No-one has actualy used a nanopore to actually sequence any length of DNA yet, so calling it 'the new way' is like saying 'the new way of driving around is in matter-antimatter cars'.

This technique is exciting if way more primitive stage than the tone of the frikkin Wired article suggests, and george church is a pioneer in sequencing, but his continuous breathless drumbeat on this topic makes me think of the many companies he's founded and continues to try and raise money for and the stockprices of.
posted by lalochezia at 11:09 AM on October 19, 2009 [3 favorites]

i'm not going to consider 'genetic engineering' a success until i have my flying pink elephant.
posted by sexyrobot at 11:28 AM on October 19, 2009

the new way of driving around is in matter-antimatter cars

We'd totally be driving matter-antimatter cars if GM hadn't killed them.
posted by blue_beetle at 11:29 AM on October 19, 2009 [1 favorite]

Nice to see IBM getting back on that genetic data horse _HAMBURGER
posted by Sys Rq at 11:39 AM on October 19, 2009 [1 favorite]

It strikes me that this is an area that will be rife with fraud. If you send away a DNA sample and a check for $1000, and receive a CD containing a hundred million A's, C's, G's, and T's -- how would you know if they were just randomly created by a computer? I sure wouldn't.
posted by Chocolate Pickle at 11:40 AM on October 19, 2009 [1 favorite]

Gattaca, here we come!

No, I kid (mostly). This is awesome! I'd totally shell out $1000 to have my genome sequenced. I don't know what I would do with it. It would seem a bit pretentious to print it out and wallpaper my apartment with it...

On a more serious note, this will vastly accelerate our discovery of genetic diseases and the cures for them, so this is great.
posted by Salvor Hardin at 11:42 AM on October 19, 2009

how would you know if they were just randomly created by a computer? I sure wouldn't.

Just synthesize the DNA, clone it, and see if the kid grows up to be just like you.
posted by Salvor Hardin at 11:43 AM on October 19, 2009

If you send away a DNA sample and a check for $1000, and receive a CD containing a hundred million A's, C's, G's, and T's -- how would you know if they were just randomly created by a computer? I sure wouldn't.

By sending the DNA sample to someone else and seeing whether the results they come back with are similar. Or relying on the idea that others are doing the same, and that it would be a business-ending scandal associated with criminal charges and ended careers if it was to be found that someone was just making up the results.
posted by grouse at 11:43 AM on October 19, 2009 [4 favorites]

Or relying on the idea that others are doing the same, and that it would be a business-ending scandal associated with criminal charges and ended careers if it was to be found that someone was just making up the results.

lol
posted by DU at 11:47 AM on October 19, 2009 [5 favorites]

It would seem a bit pretentious to print it out and wallpaper my apartment with it...

You could accomplish the same thing by exploding.
posted by Faint of Butt at 11:49 AM on October 19, 2009 [4 favorites]

Yeah, go ahead and laugh, DU, but the fact is that, contra Chocolate Pickle, it would be so ridiculously easy to detect this kind of fraud, that you'd have to be an idiot to think about doing it.
posted by grouse at 11:51 AM on October 19, 2009 [5 favorites]

Why pay $1000, when that arrives, if it's falling at 10x annually? Just wait a couple years. Even if it's just a factor of 4 a year, a few years should do it. I wonder this about science too, when it's actually worth the millions-dollar NSF grant to do sequencing or computer work now, rather than just waiting a few years and doing it on a desktop. Some of that research is necessary for advancing the curve, and some is necessary to save lives today that will be gone tomorrow; but it seems like a lot of it could wait -- freeing up lots of resources to use on projects that won't get much cheaper -- were it not for the fact that scientific glory goes only to the first.
posted by chortly at 11:52 AM on October 19, 2009 [1 favorite]

Genome Wallpaper (in 2005)
posted by blue_beetle at 12:17 PM on October 19, 2009

Gattaca, here we come!

You might be interested to read about the Genetic Information Nondiscrimination Act of 2008.
posted by Rhomboid at 12:24 PM on October 19, 2009

This could be revolutionary, in a number of different ways:

1) The current DNA fingerprinting, which uses maybe thirteen loci (I think this is up from five when I first heard about it) goes from "it's superduper unlikely" to "you would get struck by lightning while holding a winning lottery ticket, just before a meteor knocked it out of your hand" levels of unlikely.

2) Average joes and janes go berserk looking for lethal alleles (what is it, three to seven for every person, on average) and single nucleotide polymorphisms, self-diagnose themselves with any number of things, while missing out on the Big Picture.

3) Genealogy nuts worldwide begin to use mitochondrial DNA and some carefully selected subset of plain ole nuclear DNA as part of their profiles, for tracking — the leap from names to genetic data would be as powerful as the rebuilding of biological taxonomies by evolution over the older fashion of using phenotype.

4) Tailored medicine might pin down profiles of individuals who do or do not respond to particular medications well.

After we extend past SNPs, we'll see your basic Gattaca happen. It's a tool nobody would put down. Imagine if ten dollars would get you a readout on health problems on the guy you've gone on five dates with — grab a few hairs. Insurance companies everywhere quiver perpetually on the edge of orgasm at the thought of health assessments three decades hence. Peeing in a cup will seem as quaint as having three references on your resume. Not a lot of dopamine-inhibiting receptors? We want someone solid and reliable for this job.

In a grimly hilarious fashion, racism will probably evaporate to be replaced with genetic discrimination, only this time we might have some science to back it up. No longer will you not be hired due to the color of your skin at birth, but because of the fact that, also at birth, you were predisposed to have an enormous coronary in your fifties.

Lest I be labeled a Luddite, my suggestion is that we take up this tool with some form of restraint, legal or otherwise, and that it must have fast on its heels some form of genetic remediation for those born at a perceived disadvantage. Otherwise, we will create a brand new underclass; I'm afraid market forces and human nature will not have it otherwise.
posted by adipocere at 12:25 PM on October 19, 2009 [2 favorites]

I wonder this about science too, when it's actually worth the millions-dollar NSF grant to do sequencing or computer work now, rather than just waiting a few years and doing it on a desktop.

I'm posting this comment retroactively from my Higgs-based iPhone Yotta in the year 2027. The rapacious Genelords (may their line endure forever!) turn out some pretty cool tech, though sometime I wish they wouldn't stomp their boots in my face quite so often.
posted by Iridic at 12:28 PM on October 19, 2009 [1 favorite]

Just synthesize the DNA, clone it, and see if the kid grows up to be just like you.

I wonder how long it'll be before sequencing a strand of DNA is cheaper than synthesizing it.* Are we there already? DNA synthesis was pretty expensive just a couple years ago - if I recall the reason why, it's that it's exponentially hard to make stable chains that won't break down.

* I don't mean to suggest that the two are in any way fungible. I know it sounds like I'm suggesting that. But I'm not, and they are not. They are, if anything, complementary.
posted by jock@law at 12:49 PM on October 19, 2009

> I wonder how long it'll be before sequencing a strand of DNA is cheaper than synthesizing it.* Are we there already?

Always has been. I can get 1kb of DNA sequenced for about $10; I'm guessing a couple thousand for synthesis of same.
posted by nowonmai at 12:57 PM on October 19, 2009 [2 favorites]

Why pay $1000, when that arrives, if it's falling at 10x annually? Just wait a couple years. Even if it's just a factor of 4 a year, a few years should do it. I wonder this about science too, when it's actually worth the millions-dollar NSF grant to do sequencing or computer work now, rather than just waiting a few years and doing it on a desktop. Some of that research is necessary for advancing the curve, and some is necessary to save lives today that will be gone tomorrow; but it seems like a lot of it could wait -- freeing up lots of resources to use on projects that won't get much cheaper -- were it not for the fact that scientific glory goes only to the first.
posted by chortly at 11:52 AM on October 19 [1 favorite +] [!]

Seriously? SERIOUSLY? Do you not get the fact that the only reason the price for something comes down is because there is a *something* there to change and make cheaper? It's not like innovation from nothing is going to be cheaper in 10 years. It's that each stage of refinement is cheaper and cheaper.
posted by amelioration at 1:01 PM on October 19, 2009 [6 favorites]

What amelioration says: the reason these bigger faster cheaper machines are being invented is because there is a competitive market for them right now. Researchers want all the data they can get, and this market is what is driving the price down. If research money wasn't being poured into this industry right now, we'd still be relying on grad students running radioactive acrylamide slab gels.
Is it worth individuals paying for their own genomes now, or at $1000? I don't think so, because there is no obvious utility in having the data. However, for a hypochondria hobbyist, it could be a worthwhile investment. If early adopters didn't pay $4000+ for home computers 25 years ago would we have $200 netbooks and iPhones now? Somebody has to provide revenue for the companies developing this stuff while we wait for our $10 genomes to arrive.
posted by nowonmai at 1:15 PM on October 19, 2009 [2 favorites]

MetaFilter: still relying on grad students running radioactive acrylamide slab gels.
posted by Sys Rq at 1:17 PM on October 19, 2009 [2 favorites]

MetaFilter: still relying on grad students running radioactive acrylamide slab gels.

No wonder the site feels slow today.
posted by jock@law at 1:20 PM on October 19, 2009 [2 favorites]

I was trying to make that clear, but perhaps I failed. It seems logically unlikely to be that case that *every* research project utilizing technology X is necessary for the advancement of that technology X. For instance, in social science or medicine, whether you do a project that costs a million dollars of computer time now, or a thousand dollars of computer time in 5 years, is unlikely to have an effect on the advancement of Moore's law under any assumptions other than a hand-wavy, "it's all connected" kind of way. Obviously, lots of other research will have direct and indirect impacts on the technology used in that research. But it seems evident that at least some --- and probably quite a lot -- of research using rapidly advancing technology is not actually necessary for the advancement of that technology. Gene sequences, for example, are not playing much of a role in IBM's chip design, except insofar as they are providing business; much NIH or social science research is similarly not advancing computer design. It is often taken as an article of faith that any expense is worth it, but it's a matter of opportunity cost: if the NSF or NIH funds the a few projects 5 years later for 1 thousandth of the cost, that's a lot of other projects that can be funded now -- projects that in fact may advance the development of computers or sequencing faster, if that is your primary goal. It is not at all clear that rerouting a percentage of funding away from costly-now/cheap-later projects towards other things will break or even slow the development of rapidly advancing technologies.
posted by chortly at 1:24 PM on October 19, 2009

25 years ago the future of genetics still featured manual typewriters.
posted by gimonca at 1:27 PM on October 19, 2009

Sorry: my previous post is responding to amelioration above, if that wasn't clear:

Seriously? SERIOUSLY? Do you not get the fact that the only reason the price for something comes down is because there is a *something* there to change and make cheaper? It's not like innovation from nothing is going to be cheaper in 10 years. It's that each stage of refinement is cheaper and cheaper.

This seems to suppose that I was suggesting defunding all science and just waiting, which is silly. I was suggesting that there probably exists a subset of projects that utilize costly and rapidly advancing technologies which could have their funding deferred until it's cheaper, with the savings spent on other research. Arguing that any such change would break the engine of progress is a bit of a stretch, it seems to me.
posted by chortly at 1:29 PM on October 19, 2009

This is medical research. If I'm researching a genetic disease, call it pancakeitis, I want the genome sequences of as many pancakeitis patients as I can afford. If I can get 100 today, that's a great start; 1000 in 2 years time would be even better, but I'm not going to sit on my ass and let those jackasses at Harvard get a head start and all the glory. That's how research works; you do what you can NOW because somebody else will do it if you don't. Of course NIH et al. take bang-for-the-buck into consideration when they award grants - right now they clearly think large-scale sequencing projects are already providing value for money compared to the other projects being pitched.
The reason IBM is in this is not a blue sky technology project hoping for spin-off benefits; if their machine produces 5 times as much sequence per day as Illumina's, they can sell a lot of them at $100,000 a pop. And at the high end, this research IS pushing the limits of computing performance.
posted by nowonmai at 1:38 PM on October 19, 2009 [2 favorites]

No-one has actualy used a nanopore to actually sequence any length of DNA yet, so calling it 'the new way' is like saying 'the new way of driving around is in matter-antimatter cars'.

You said it, man. I've spent awhile working with this technology, and I'd be surprised if it turns out to be viable for sequencing. No one has been able to say what the signal would be yet. How does the nanopore distinguish bases?
posted by mr_roboto at 1:39 PM on October 19, 2009

Too bad the concept is bullshit: Mythbusting Personalized Genomics
posted by blasdelf at 2:04 PM on October 19, 2009

if the NSF or NIH funds the a few projects 5 years later for 1 thousandth of the cost, that's a lot of other projects that can be funded now -- projects that in fact may advance the development of computers or sequencing faster, if that is your primary goal

There's also a huge amount of small-scale use of these technologies, and I think you're ignoring the incremental impact of these projects beyond the simple impact of improving computing and sequencing.

My point is that we don't have to pick between "now" and "some future time when the technology is cheap and all the boring questions can get to use it." PIs are generally very careful about how they spend their funding, and they exhaust the cheap tech options before moving on to the big new shiny technology. But when they do decide to use that big new shiny technology, they're helping to create the drive that pushes it forward. Besides which, they're spending that money not just to advance big new shiny technology on not-directly-relevant questions, they're answering very important questions in their own field.
posted by amelioration at 2:08 PM on October 19, 2009

Too bad the concept is bullshit: Mythbusting Personalized Genomics

That link has nothing to do with what is being discussed here. It is correct in that current "personalized genomics" technology does not involve full sequencing. This post, however, is about a new technology that is intended to facilitate full sequencing.
posted by mr_roboto at 2:25 PM on October 19, 2009 [4 favorites]

One of the major obstacles with sequencing DNA using nanopores is that current photolithography techniques aren't quite able to pattern small enough pores reliably. The width of double stranded DNA is 2.4 nanometers and current technologies for creating pores crap out around 10 nanometers or so (I forget the exact number).

Interestingly enough, one way to achieve precise alignment at these scales may be to use self assembling DNA itself as a template.

Right now the most advanced DNA sequencing methods work on single molecules already and some don't require amplification.
posted by euphorb at 2:28 PM on October 19, 2009 [1 favorite]

But when they do decide to use that big new shiny technology, they're helping to create the drive that pushes it forward. Besides which, they're spending that money not just to advance big new shiny technology on not-directly-relevant questions, they're answering very important questions in their own field.

That's how research works; you do what you can NOW because somebody else will do it if you don't.

I guess I'm thinking less from the researcher's point of view, who as I said above, clearly has a professional incentive to be first, and more from the point of view of the funding institution, or societal benefit maximization more broadly. If, as a funder, I can buy either project A now for X dollars, or I can buy B,C...H projects now plus project A in five years all for X dollars, why not go with the latter? It's true that in some cases, this delay of A will stunt the cost decline I am relying on -- but probably not for most projects. If the same question gets answered at a fraction of the cost in five years rather than today, in many cases the delay will be worth it in order to fund all those other projects which otherwise would not have existed at all. And even from the self-interested point of view of scientists, more scientists will be made happy under the latter scenario (though of course not the project A folks).
posted by chortly at 2:36 PM on October 19, 2009

Well, the researcher's professional incentive to be first comes from the fact that if you're not first, you probably won't get to do it at all -- after all, redundant research is even harder to justify funding for than expensive innovation.

As a funder, though, you wouldn't just be looking at a single impact of project A. The decision to fund any project is a complex one, involving not only the cost, but also the impact of the product(s) of that project on fields X, Y and Z, and also the likelihood of success (which is, in turn, impacted by the technologies available to the researcher). If all is equal but the cost, then the cheaper projects are generally more likely to get funded (or at least, so reports my advisor after he returns from participating in funding panels). There are a huge number of other intricacies to the federal funding process and scientific progress that are immensely discussion-worthy, but I think that conversation is a bit of a derail to the discussion of IBM's newest foray into the world of genomics technology, as well as to the potential costs and benefits of full-scale personalized genome sequencing.
posted by amelioration at 3:00 PM on October 19, 2009 [1 favorite]

I can't be the only person to read this and then be unable to stop singing "Your Own...Personal....Genome..."
posted by ilana at 3:22 PM on October 19, 2009

> I can't be the only person to read this and then be unable to stop singing "Your Own...Personal....Genome..."
You can't imagine how profoundly I hate you right now.
posted by nowonmai at 3:34 PM on October 19, 2009 [1 favorite]

Rhomboid: Gene based discrimination was illegal in Gattica, too.
posted by absalom at 3:46 PM on October 19, 2009

There are a huge number of other intricacies to the federal funding process and scientific progress that are immensely discussion-worthy, but I think that conversation is a bit of a derail to the discussion of IBM's newest foray into the world of genomics technology, as well as to the potential costs and benefits of full-scale personalized genome sequencing.

Point taken. I was mostly just defending my fairly anodyne comment from the "Seriously? SERIOUSLY?" reaction.
posted by chortly at 4:21 PM on October 19, 2009

Further, it's worth pointing out that scientific results have accelerating returns; results from research this year allows further research next year, which allows even more research the year following. Particularly in a field this new, where we know so little, this is very much like compound interest, and spending $100,000 this year to spark off $10,000,000 in research next year may be completely worth it, even if waiting a year drops the initial $100k to $10k.

I don't know if this is still holds true, but Asimov pointed out late in his career that, over the broad span of history, spending on research, especially basic research, typically paid off at about 10 to 1. This isn't because every research project pays off; rather, it's because occasionally, you get something like the transistor that pays off millions to one, and those really big advances need all the other "dead-end" projects to provide data to make the big insight. Looking at individual projects, much research looks like wasted money, but taken in as a whole, it pays off BIG, probably better than any other field of human endeavor.

So getting those results done and out there this year is important; it accelerates the whole research process, and brings the (inevitable?) huge payoffs that much closer.

Also, one more thing to keep in mind is that even if the hardware drops in cost that fast, the people that RUN the hardware are the real treasure, and we need to keep their minds busy. Those super-smart people are very, very rare. They're the part of research that really matters, and every year they're deprived of data to work with is a productive year of their lives gone forever.
posted by Malor at 4:57 PM on October 19, 2009 [3 favorites]

OT: So many posts, and nobody complimented the OP for the Bowie reference? Well done, sir/madam.

Carry on.
posted by VikingSword at 6:25 PM on October 19, 2009

every year they're deprived of data to work with is a productive year of their lives gone forever

The field of genomics does not suffer for lack of data. In fact, data production has more or less followed an exponential trend since the early 1990s, well outstripping the pace of analysis.
posted by Blazecock Pileon at 6:34 PM on October 19, 2009

It strikes me that this is an area that will be rife with fraud. If you send away a DNA sample and a check for $1000, and receive a CD containing a hundred million A's, C's, G's, and T's -- how would you know if they were just randomly created by a computer? I sure wouldn't.

If you pick some subsequences at random and search for them against a reference genome assembly, you'll find out quickly.
posted by Blazecock Pileon at 6:50 PM on October 19, 2009

Give it up, IBM. No self-respecting gnome would work for a mere thousand dollars.
posted by kid ichorous at 8:10 PM on October 19, 2009 [1 favorite]

NHGRI Awards $1,000 Genome Grants to IBM Research, Ion Torrent, Others: lists other companies and academics who are doing similar work.

A lot of smart people in diverse areas have been working on nanopore sequencing for close to a decade. This story stinks like it was crafted by the IBM PR department.

The Genographic Project is very cool, however.
posted by benzenedream at 12:04 AM on October 20, 2009

So what IBM can actually do at the moment is drill a tiny hole using an electron microscope, then push single-stranded DNA through using an electric voltage, with periodic pauses. They can't actually read the sequence of the strand, they can just move it and stop it. While that was undoubtedly harder than it sounds, they really don't have a sequencing method here.

This sounds like PR puffery. The hardest part - reading the sequence - is still a ways off and will probably be orders of magnitude more difficult than this.

And concerning the myth of "read a DNA sequence, predict somebody's medical future", it kinda doesn't matter that No You Can't. Human nature being what it is, HR departments and insurance companies will use genetic information to "optimize" hiring practices and set premiums. Probably without actually admitting to it, either. If there are 2 candidates for a job and one has genes that correlate with high risk of cancer/heart attack/other-debilitating-and-expensive-condition, guess who's going to get the job? The other candidate, who is better qualified and has more appropriate experience in this job sector. Correlation, causation, who gives a shit - there's always tons of candidates for any given job and we don't have to explain why we didn't hire any of the people we didn't hire, and who wants to take a risk on somebody who might go on disability? Insurance companies have all kinds of arcane formulas for setting premiums; throw in a few more factors and who's gonna notice?

I mean, racism still hasn't gone away and that's the most baseless form of discrimination. If someone has an amplified HER2/neu gene, which correlates with more aggressive breast cancer, there's a scientific basis for not touching this candidate/client with a ten-foot pole. There's just too much financial incentive to use genetic data to (try to) screen out expensive/problematic people, and I'm cynical about legal safeguards (see: racism). Again, it doesn't matter if DNA sequences don't actually tell you that - people will believe they're stacking the deck in their favor, and confirmation bias will take over from there. (See? Larry still hasn't had a heart attack - told you he has good genes!)
posted by Quietgal at 9:28 AM on October 20, 2009 [2 favorites]