master of information
April 9, 2011 11:00 AM Subscribe
Nice tag list
posted by Jumpin Jack Flash at 11:28 AM on April 9, 2011 [3 favorites]
posted by Jumpin Jack Flash at 11:28 AM on April 9, 2011 [3 favorites]
As the guy who is trying to sell you the newest next-gen sequencing platform, of course, he is trying to get everyone to embrace complexity. But what are his solutions for dealing with this complexity? Networks? Systems Biology? More GWAS? These are buzzwords, and none of them are new. I get the feeling that Erick Schadt is an amazing salesman, but what's he selling? More machines to do more sequencing. It's a great evolutionary step to having more things to study, but how that will upend molecular biology and open source it, I'm not getting from this article.
posted by reformedjerk at 11:40 AM on April 9, 2011 [5 favorites]
posted by reformedjerk at 11:40 AM on April 9, 2011 [5 favorites]
altho written in a slightly annoying style imo, i guess the two things it brought out for me was the apparent interpretive impasse that gene sequencing has foundered on and which schadt is attempting to work thru; i think charlie stross put it best:
---
*it seems, like big banks, "they're happy letting startups do innovation and then copying what works"
posted by kliuless at 11:42 AM on April 9, 2011 [3 favorites]
There's been enormous progress in genomics; we're now on the threshold of truly understanding how little we understand. While the anticipated firehose of genome-based treatments hasn't materialized, we now know why it hasn't materialized, and it's possible to start filling in the gaps in the map. Turns out that sequencing the human genome was merely the start. (It's not a blueprint; it's not even an algorithm for generating a human being. Rather, it's like a snapshot of the static data structures embedded in an executing process. Debug that.) My bet is that we're going to have to wait another decade. Then things are going to start to get very strange in medicine.and the other was the extent that big pharma -- a confusopoly -- isn't interested:*
...he and Friend tried to turn Merck into a New Biology company, by which they meant a company that would share its data with networks of outside scientists and that would develop drugs that targeted networks instead of single genes. The problem with that: Merck was still an old biology company. The drugs in its pipeline — including the drugs informed by Schadt's networks — targeted single genes. And so when Schadt and Friend made their presentation, this was Merck's response: "We're not an information company." And when, in 2009, Schadt published a paper in Nature entitled "A Network View of Disease and Compound Screening" — a paper that implied that drugs targeting single genes were doomed to failure — "well, that was the paper that got me kicked out of Merck."and so, given the great stagnation in medicine, i'm pretty partial to:
Baker's alternative to all the distortions created by the patent system is simple direct public funding for medical R&D and education: the costs of such a scheme would be lower than the amount we're already paying on publicly-funded prescription drug benefits, so it would save money on a fiscal basis. All patents and research results would go into the public domain, which would generate huge global benefits.afterall, public sector drug discovery has been rather productive.
---
*it seems, like big banks, "they're happy letting startups do innovation and then copying what works"
posted by kliuless at 11:42 AM on April 9, 2011 [3 favorites]
It's a great evolutionary step to having more things to study, but how that will upend molecular biology and open source it, I'm not getting from this article.
I think you have the core of it. This article is written as if molecular biologists think that they're satisfied with the current reductionist approach; it implies scientists have looked at the "one gene one protein" idea, say back, and stopped what they were doing.
In reality, Schadt's advocacy seems like a loud vocalization of everyone's subtext: we're here, doing the best we can do, until the technology gets cheaper and the techniques get better, then we'll keep gradually scaling up.
I recently read a paper in which TWO novel ant genomes were sequenced just so they could be compared. Hell even organismal biologists, natural history guys, heavily rely on PCR to get at the sort of big-scale effect Schadt's "selling." Are these not indicators of a progression to large scale genetics?
Molecular biology has not failed.
posted by Buckt at 11:55 AM on April 9, 2011 [4 favorites]
I think you have the core of it. This article is written as if molecular biologists think that they're satisfied with the current reductionist approach; it implies scientists have looked at the "one gene one protein" idea, say back, and stopped what they were doing.
In reality, Schadt's advocacy seems like a loud vocalization of everyone's subtext: we're here, doing the best we can do, until the technology gets cheaper and the techniques get better, then we'll keep gradually scaling up.
I recently read a paper in which TWO novel ant genomes were sequenced just so they could be compared. Hell even organismal biologists, natural history guys, heavily rely on PCR to get at the sort of big-scale effect Schadt's "selling." Are these not indicators of a progression to large scale genetics?
Molecular biology has not failed.
posted by Buckt at 11:55 AM on April 9, 2011 [4 favorites]
Goddamnit, why does science journalism need to be so terrible? This breathlessly shitty article goes through factual errors, gross misrepresentations, and Pepsi Blue-esque bullshit like truth were a box of kleenex in 1918.
This guy is not throwing away molecular biology, he is simply advocating a genomic approach, a science that has been around for years, while the author disparages genetics, its predecessor. It makes sense since he is making one of the many possible fancy new machines for genomic sequencing, and honestly, these guys will probably do it better. (his method being limited by a need for a constant high concentrations of exotic nucleotides that have got to be damn expensive to make, the accuracy of the polymerase that must also accept his nucleotides, and the accuracy of the detector)
"Yes, he's already glimpsed some things with the PacBio RS — he's looking to prove that instead of four bases making up the DNA molecule, there are actually so many modifications of the four that the real number could be more than twenty. (It's a perfect Eric Schadt breakthrough, because not only would it be a "game changer," it would also complicate the practice of biology beyond human capability.)"
We've known this since the fifties. There were already more than 20 modified or substituted non-cononical bases described when this guy was born.
"He's cofounder of a nonprofit organization called Sage Bionetworks, which is dedicated to facilitating biological research through an open-source sharing of data."
Sage Bionetworks does awesome shit, but genomic research is already open-source, seriously check it out, all but the newest sequenced genomes are available for your perusal, right now, indexed for the world courtesy of Uncle Sam.
posted by Blasdelb at 12:10 PM on April 9, 2011 [8 favorites]
This guy is not throwing away molecular biology, he is simply advocating a genomic approach, a science that has been around for years, while the author disparages genetics, its predecessor. It makes sense since he is making one of the many possible fancy new machines for genomic sequencing, and honestly, these guys will probably do it better. (his method being limited by a need for a constant high concentrations of exotic nucleotides that have got to be damn expensive to make, the accuracy of the polymerase that must also accept his nucleotides, and the accuracy of the detector)
"Yes, he's already glimpsed some things with the PacBio RS — he's looking to prove that instead of four bases making up the DNA molecule, there are actually so many modifications of the four that the real number could be more than twenty. (It's a perfect Eric Schadt breakthrough, because not only would it be a "game changer," it would also complicate the practice of biology beyond human capability.)"
We've known this since the fifties. There were already more than 20 modified or substituted non-cononical bases described when this guy was born.
"He's cofounder of a nonprofit organization called Sage Bionetworks, which is dedicated to facilitating biological research through an open-source sharing of data."
Sage Bionetworks does awesome shit, but genomic research is already open-source, seriously check it out, all but the newest sequenced genomes are available for your perusal, right now, indexed for the world courtesy of Uncle Sam.
posted by Blasdelb at 12:10 PM on April 9, 2011 [8 favorites]
There weren't many details of what seems to be Schadt's underlying views about the epistemology (and metaphysics) of biology, but it certainly seems like he's calling for a modification (if not an outright rejection) of a reductionist program. The reductionist temptation had been to find explanations for the entire workings of a biological entity by understanding the organization, function and relation of some set of basic components one to another. Given the lack of progress that Schadt thinks were making on basic molecular biology questions, it seems plausible that reductionism needs to be rethought, but what paradigm could supplant it? The "New Biology" is about networks and how their behavior may have causal interaction with the basic building blocks of those networks -- the sounds like a loose analogue of what philosophers of mind call "downward mental causation". Which I've always thought of as a sort of relabelling of mind/body problem with an added implication that the problem's too hard to solve. How could we understand the complexities of the feedback loops involved with both upward (which could be understood reductively) and downward causation. Maybe the general outline is the right one for understanding the difficulty of the issue and the general structure of the how a solution might eventually be found, but there's not a very good road map for how to start heading toward that solution.
I'm curious about how progress in an such an approach to molecular biology would go. Can those involved in the New Biology find laws governing the processes that they believe to be key to biology in an analytical fashion or must the progress be made with the brute force of computation and statistical modelling?
posted by Wash Jones at 12:11 PM on April 9, 2011
I'm curious about how progress in an such an approach to molecular biology would go. Can those involved in the New Biology find laws governing the processes that they believe to be key to biology in an analytical fashion or must the progress be made with the brute force of computation and statistical modelling?
posted by Wash Jones at 12:11 PM on April 9, 2011
Wash Jones, the epistemological shift being described happened more than thirty years ago.
posted by Blasdelb at 12:18 PM on April 9, 2011
posted by Blasdelb at 12:18 PM on April 9, 2011
Blasdelb, Thanks for the link. I don't know much about this topic. What is Schadt offering that's new and innovative?
posted by Wash Jones at 12:24 PM on April 9, 2011
posted by Wash Jones at 12:24 PM on April 9, 2011
I doubt Schadt's worst enemies acting in concert could make him look as bad as that article does.
posted by jamjam at 12:28 PM on April 9, 2011 [3 favorites]
posted by jamjam at 12:28 PM on April 9, 2011 [3 favorites]
breathless is a perfect description of that article.
posted by a womble is an active kind of sloth at 12:33 PM on April 9, 2011
posted by a womble is an active kind of sloth at 12:33 PM on April 9, 2011
Before actually reading: Why are there so many pictures of this guy's thighs? Gah. No one needs to see that under the desk angle.
posted by maryr at 12:36 PM on April 9, 2011 [3 favorites]
posted by maryr at 12:36 PM on April 9, 2011 [3 favorites]
From the descriptions, I felt certain that this would be a Wired article.
posted by leotrotsky at 12:42 PM on April 9, 2011 [2 favorites]
posted by leotrotsky at 12:42 PM on April 9, 2011 [2 favorites]
I get the feeling that Erick Schadt is an amazing salesman, but what's he selling?Schadt's papers are seminal advances. Another name here is Trey Ideker, but looking over both their publication records over the last decade, even Ideker's papers (which are always impressive) look simpler and more ad hoc than Schadt's completely original approaches. Of course, Schadt has had many more resources put at his fingertips, which is hugely important for an expensive science like biology.
Schadt is the real deal when it comes to modeling complex systems, and verifying the predictions of his models. His work a few years ago on the communication between multiple organs, as well as the communications within each cell within each organ, was truly astounding. At the time I had been hoping to do something like that but knew I'd have to wait 5-10 years for the data, and here Schadt had plowed ahead and even incorporated more data types than I had even hoped for.
Sure, it's obvious that we need to make a synthesis of all the interactions of genes, but Schadt is the one making it happen, and the guy getting pushback from older molecular biologists and institutions like Merck. I still question every person I meet from Merck on Schadt's departure, because I really can't believe that Merck would let such a talent go. The sad conclusion from these conversations is that it really does sound like Merck just didn't understand the talent they had. And that should scare us all; if corporations as greedy as big pharma are unwilling to pick up money sitting on the table because they don't want to be an information company, then they've fundamentally misunderstood the modern world and modern biology and their place in it. How many cures and advances aren't going to make it to people because of silly clinging to old methodologies?
The failure of all these GWAS studies is obvious and predictable once you realize they're not even paying attention to cellular circuitry. But what gets funded year after year? Another million dollars here to find low-confidence, rarely reproducible single-gene "causes." GWAS is the epitome of old-style molecular biology ignoring what should be obvious by now. One must think of the whole system when trying to understand complex diseases. The GWAS community is really hoping that looking at whole-genome resequencing will be able to find single-gene causes in the rare, 1%-5% prevalance allelles, another huge jump in expense for an unlikely benefit. They're willing to adopt new technology for getting more of the same type of data, but they're unwilling to make the conceptual leap to thinking of the cellular network, beyond simple and hopelessly underpowered epistasis tests. Even testing a simple "bag of genes" is considered advanced, and always done half-heartedly.
Eric Schadt has been one of my biggest intellectual influences, and I really think he's pioneering the way forward in biology. I think that in 20 years we'll look back and see that he's the Jacques Monod of our century, slowly building a body of work that leads to large and general insights that will talke another half-century to begin to fully understand. Sure, I'm breathless, but I really do find his work exciting. (But it's always bittersweet, because I know he's going to do some of things I want to do first.)
posted by Llama-Lime at 12:43 PM on April 9, 2011 [12 favorites]
I didn't think that it would have been possible to write an article about Systems Biology without using that term at all but apparently it is
The idea that biology could be understood through biochemistry (or molecular biology) alone is pretty dated at this point. But now that computational and sequencing tools are becoming more widespread, different kinds of questions can be asked and answered.
It makes sense since he is making one of the many possible fancy new machines for genomic sequencing, and honestly, these guys will probably do it better.
Maybe nanopore sequencing will be the greatest thing in the world down the road but right now the best it has to offer is a 10+ year old trail of publications that have been hyped to death but no sequencers on the market currently. Meanwhile Pacific Biosciences' main technology is only 8 years old and they have working sequencers for sale now.
Right now I think Ion Torrent has the technology to beat. The instrument is less expensive to buy and less expensive to run. But this is an area that moves so fast that new technologies can quickly make older ones look like dinosaurs.
posted by euphorb at 1:06 PM on April 9, 2011
The idea that biology could be understood through biochemistry (or molecular biology) alone is pretty dated at this point. But now that computational and sequencing tools are becoming more widespread, different kinds of questions can be asked and answered.
It makes sense since he is making one of the many possible fancy new machines for genomic sequencing, and honestly, these guys will probably do it better.
Maybe nanopore sequencing will be the greatest thing in the world down the road but right now the best it has to offer is a 10+ year old trail of publications that have been hyped to death but no sequencers on the market currently. Meanwhile Pacific Biosciences' main technology is only 8 years old and they have working sequencers for sale now.
Right now I think Ion Torrent has the technology to beat. The instrument is less expensive to buy and less expensive to run. But this is an area that moves so fast that new technologies can quickly make older ones look like dinosaurs.
posted by euphorb at 1:06 PM on April 9, 2011
Llama-Lime: I would be interested to hear what you think about the work of Sean B. Carroll and Richard Lewontin? As someone with an interest in evo-devo, developmental systems theory and the theoretical framework put forward by Susan Oyama, I'm curious how these ideas are faring in biology now? No snark or axe to grind, just curious.
posted by The Emperor of Ice Cream at 1:10 PM on April 9, 2011 [1 favorite]
posted by The Emperor of Ice Cream at 1:10 PM on April 9, 2011 [1 favorite]
"Blasdelb, Thanks for the link. I don't know much about this topic. What is Schadt offering that's new and innovative?"
As much as I despise the cults of personality that can develop around leading scientists, worship being such an easy replacement for understanding, Schadt is actually brilliant.
It is important to note that people who understand both computing/mathematics and biology are both rare and incredibly valuable. (The old physics joke about biologists just being physicists who can't do math is a little true, but seriously, if you want to push your kids towards a job that will make them worth their weight in gold this is how. Schadt is as invaluable as he is because he has demonstrated an ability to both produce amazingly valuable research from this combination while understanding business, he really is unique.
Back in the late 90s he and a few other people founded Rosetta Inpharmatics, a company that used a custom built massive supercomputer and a lot of talent to do some really awesome bio-informatics. The tiny startup ended up being so promising that Merck was willing to pay $620 million for it. However, it ended up not having the commercial potential that they thought it did in the way that they were hoping and they ended up disbanding it, donating most of the division to Sage Bionetworks, a non-profit built to keep the research alive. The "open-source" stuff in the article comes from this non-profit letting other people use their expertise and computing power. For Merck to just throw him and his company away really was idiotic, they hoped that with the Rosetta platform they could make their current drug discovery process more efficient, and it did but not $620 million more efficient. However, Merck wasn't able to grasp the power that really well done bio-informatics can add to genetics.
To really leverage what our understanding of molecular biology can do for our understanding of disease we need to do two things. Invest in the dying science of molecular genetics, which this article mistakenly attacks, continuing the difficult and time consuming task of identifying and characterizing new genes. While also pushing as many kinds as we can into both math and biology to do the kinds of innovation that Schadt has been publishing.
"Before actually reading: Why are there so many pictures of this guy's thighs? Gah. No one needs to see that under the desk angle."
Its actually one of the things he is best known for... The khaki shorts with the white polo anyway.
posted by Blasdelb at 1:24 PM on April 9, 2011
As much as I despise the cults of personality that can develop around leading scientists, worship being such an easy replacement for understanding, Schadt is actually brilliant.
It is important to note that people who understand both computing/mathematics and biology are both rare and incredibly valuable. (The old physics joke about biologists just being physicists who can't do math is a little true, but seriously, if you want to push your kids towards a job that will make them worth their weight in gold this is how. Schadt is as invaluable as he is because he has demonstrated an ability to both produce amazingly valuable research from this combination while understanding business, he really is unique.
Back in the late 90s he and a few other people founded Rosetta Inpharmatics, a company that used a custom built massive supercomputer and a lot of talent to do some really awesome bio-informatics. The tiny startup ended up being so promising that Merck was willing to pay $620 million for it. However, it ended up not having the commercial potential that they thought it did in the way that they were hoping and they ended up disbanding it, donating most of the division to Sage Bionetworks, a non-profit built to keep the research alive. The "open-source" stuff in the article comes from this non-profit letting other people use their expertise and computing power. For Merck to just throw him and his company away really was idiotic, they hoped that with the Rosetta platform they could make their current drug discovery process more efficient, and it did but not $620 million more efficient. However, Merck wasn't able to grasp the power that really well done bio-informatics can add to genetics.
To really leverage what our understanding of molecular biology can do for our understanding of disease we need to do two things. Invest in the dying science of molecular genetics, which this article mistakenly attacks, continuing the difficult and time consuming task of identifying and characterizing new genes. While also pushing as many kinds as we can into both math and biology to do the kinds of innovation that Schadt has been publishing.
"Before actually reading: Why are there so many pictures of this guy's thighs? Gah. No one needs to see that under the desk angle."
Its actually one of the things he is best known for... The khaki shorts with the white polo anyway.
posted by Blasdelb at 1:24 PM on April 9, 2011
From the fawning language, hyperbole, and silly details lovingly plastered into this article by the writer I'm fairly certain Rasputin has been reincarnated as a biologist.
On what this guy is trying to do: we already have collaborations and they work. Its too bad that no one wants to develop drugs based on new research x, y, and z but that's a big pharma problem (not wanting to risk big bucks in a new drug rather than making a spin off of an existing one, see the immensely tiny variety in the current stock on antibiotics) not a molecular biology problem. We already have high-throughput assay and large collaborations as multiple labs disentangle related pathways together.
It's great that he's got a giant network of genetic data that he's toying with. Unfortunately if you want to prove that any of the genetic 'hits' correspond to an actual cellular response you have to track down and characterize the individual proteins, which takes months or years. High throughput screens spit out a lot of false positives and negatives.
It would be great for one intelligent, unassuming, and self-less individual to be ringmaster and drive a hive-mind of labs towards common goals. Unfortunately, I doubt that anyone with their brain on 'publish or perish' will trust this man to give anyone else but Mr. Ego first author.
posted by Slackermagee at 1:48 PM on April 9, 2011
On what this guy is trying to do: we already have collaborations and they work. Its too bad that no one wants to develop drugs based on new research x, y, and z but that's a big pharma problem (not wanting to risk big bucks in a new drug rather than making a spin off of an existing one, see the immensely tiny variety in the current stock on antibiotics) not a molecular biology problem. We already have high-throughput assay and large collaborations as multiple labs disentangle related pathways together.
It's great that he's got a giant network of genetic data that he's toying with. Unfortunately if you want to prove that any of the genetic 'hits' correspond to an actual cellular response you have to track down and characterize the individual proteins, which takes months or years. High throughput screens spit out a lot of false positives and negatives.
It would be great for one intelligent, unassuming, and self-less individual to be ringmaster and drive a hive-mind of labs towards common goals. Unfortunately, I doubt that anyone with their brain on 'publish or perish' will trust this man to give anyone else but Mr. Ego first author.
posted by Slackermagee at 1:48 PM on April 9, 2011
Full disclosure: I am a biochemist and I am never, EVER, happy with seeing a bunch of genetic maps with no IPs, co-localizations, activity assays, or corroborating evidence of any kind. This kind of biases me against the large scale, mostly blind bioinformatics this guy does.
And to reiterate what has been said before, this Grand New Revolution In Science happened around the time my advisor and his co-workers were finishing their post-docs. Or possibly even earlier.
posted by Slackermagee at 1:53 PM on April 9, 2011 [2 favorites]
And to reiterate what has been said before, this Grand New Revolution In Science happened around the time my advisor and his co-workers were finishing their post-docs. Or possibly even earlier.
posted by Slackermagee at 1:53 PM on April 9, 2011 [2 favorites]
Actually, I'm beginning to think that salesmen hawking "paradigm-shifting" tools is precisely what's killing molecular biology.
Here's one reason why I'm wary of bio-informatics: the principle of garbage-in, garbage-out. Extrapolation turns tiny errors into massively wrong conclusions. Which are hugely costly in the long run, both in terms of time and money, as people run around in circles trying to prove or disprove them.
As our tools get more and more complex, the variables multiply exponentially, which means not only are there more potential sources of error and bias, but these sources become increasingly opaque to analysis. All these fancy new machines introduce layers of epistemological distance between manipulation and induction. So our data not only sucks, it sucks in ways we can't easily understand or predict.
So we've got all this data, and a lot of sucks. What's the new plan? Let's generate system-wide principles from it! Using complex and opaque algorithms!
You'll have to forgive my wary, slit-eyed glance.
posted by dephlogisticated at 2:35 PM on April 9, 2011 [1 favorite]
Here's one reason why I'm wary of bio-informatics: the principle of garbage-in, garbage-out. Extrapolation turns tiny errors into massively wrong conclusions. Which are hugely costly in the long run, both in terms of time and money, as people run around in circles trying to prove or disprove them.
As our tools get more and more complex, the variables multiply exponentially, which means not only are there more potential sources of error and bias, but these sources become increasingly opaque to analysis. All these fancy new machines introduce layers of epistemological distance between manipulation and induction. So our data not only sucks, it sucks in ways we can't easily understand or predict.
So we've got all this data, and a lot of sucks. What's the new plan? Let's generate system-wide principles from it! Using complex and opaque algorithms!
You'll have to forgive my wary, slit-eyed glance.
posted by dephlogisticated at 2:35 PM on April 9, 2011 [1 favorite]
So my bread and butter is a polyantigen ELISA to detect host cell proteins in purified recombinant materials. It's not a well behaved assay for a lot of reasons. A few years ago this got me to thinking hard about the net effect of a couple hundred different antibody-antigen reactions all going at the same time. A mess of excel spreadsheets later and I now have a pretty intuitive grasp of the subtitles of weird linearity issues, spike recovery weirdness and so on that I've been seeing.
What I've found is that a lot of people aren't willing to embrace the idea of a system with a lot of moving pieces and little certainty, even though we've know that's how things work for 80 years or so (it's not like I discovered this stuff or anything). Instead, everyone wants to embrace a paper from the nineties that used the term "molar excess".
In retrospect, I can see why. A > B is easy. You can explain it to an auditor or some executive type in a minute or so. The summation of the intersection Γ=n(D∙t/x)^0.5 and K=[A][B]/[AB] over a couple hundred protein species and antibodies, each with their own affinity, is like the biology version of that Let's Make a Deal problem - the one where Monty opens one of his two doors to reveal some goats, and asks if you should change your guess, only with more calculus and fewer boxes that Jay is standing next too. AN FDA auditor is going to treat you like you're trying to distract him from some obvious weakness in your work or like you are nuts. Or both.
Down in the trenches I've seen people embrace hypotheses that were obviously gross oversimplifications of what was going on, but it was pretty clear that they had an easy to understand argument and were perfectly willing to spend the whole day sitting in a meeting room restating it loudly. The counter argument would involve looking up the hydrodynamic radius of a molecule, doing some math, trying to find the viscosity of Dulbecco's PBS with 1 mg/mL BSA in it, doing some more math and then getting people to follow all this math to the conclusion. And I have a report that has to get in the document system by Friday and I will be a bad person in the eyes of the company if that critical deadline is not met.
"Sounds great - let me know how your experiment turns out."
Of course Merck let this guy go. Eric Schadt was a resource they didn't know how to use - it's like all those small businesses that bought a computer in the late seventies that they then didn't know what to do with. We laugh at them now but the truth is that they had a system that didn't need a computer and they didn't have the time or inclination to change their system. Merck has a business model. It's worked in the past. If they just have the guy who turns the crank on the drug discovery box turn it a little faster everything will be fine. Heck, maybe it even will.
So yeah, all the breathless hyperbole aside, I am 100% with where this guy is coming from. The citadel of Big Pharma where I work is like a shrine to reductionist thought and full of resources that aren't being utilized to their best potential.
On the other hand, I can't really think of any other industry that really embraces complexity or systems thinking and while almost everyone I know has a computer that would put a Cray 2 to shame, more of them have a level 80 night elf hunter than a research paper on computational bio-kinetics.
References available upon request.
posted by Kid Charlemagne at 2:39 PM on April 9, 2011 [9 favorites]
What I've found is that a lot of people aren't willing to embrace the idea of a system with a lot of moving pieces and little certainty, even though we've know that's how things work for 80 years or so (it's not like I discovered this stuff or anything). Instead, everyone wants to embrace a paper from the nineties that used the term "molar excess".
In retrospect, I can see why. A > B is easy. You can explain it to an auditor or some executive type in a minute or so. The summation of the intersection Γ=n(D∙t/x)^0.5 and K=[A][B]/[AB] over a couple hundred protein species and antibodies, each with their own affinity, is like the biology version of that Let's Make a Deal problem - the one where Monty opens one of his two doors to reveal some goats, and asks if you should change your guess, only with more calculus and fewer boxes that Jay is standing next too. AN FDA auditor is going to treat you like you're trying to distract him from some obvious weakness in your work or like you are nuts. Or both.
Down in the trenches I've seen people embrace hypotheses that were obviously gross oversimplifications of what was going on, but it was pretty clear that they had an easy to understand argument and were perfectly willing to spend the whole day sitting in a meeting room restating it loudly. The counter argument would involve looking up the hydrodynamic radius of a molecule, doing some math, trying to find the viscosity of Dulbecco's PBS with 1 mg/mL BSA in it, doing some more math and then getting people to follow all this math to the conclusion. And I have a report that has to get in the document system by Friday and I will be a bad person in the eyes of the company if that critical deadline is not met.
"Sounds great - let me know how your experiment turns out."
Of course Merck let this guy go. Eric Schadt was a resource they didn't know how to use - it's like all those small businesses that bought a computer in the late seventies that they then didn't know what to do with. We laugh at them now but the truth is that they had a system that didn't need a computer and they didn't have the time or inclination to change their system. Merck has a business model. It's worked in the past. If they just have the guy who turns the crank on the drug discovery box turn it a little faster everything will be fine. Heck, maybe it even will.
So yeah, all the breathless hyperbole aside, I am 100% with where this guy is coming from. The citadel of Big Pharma where I work is like a shrine to reductionist thought and full of resources that aren't being utilized to their best potential.
On the other hand, I can't really think of any other industry that really embraces complexity or systems thinking and while almost everyone I know has a computer that would put a Cray 2 to shame, more of them have a level 80 night elf hunter than a research paper on computational bio-kinetics.
References available upon request.
posted by Kid Charlemagne at 2:39 PM on April 9, 2011 [9 favorites]
I'm fairly certain Rasputin has been reincarnated as a biologist.
If you saw my last company ID picture, you'd be certain of it.
All I could think when I first saw it was, "I don't really look this much like a druid priest, do I?"
posted by Kid Charlemagne at 2:43 PM on April 9, 2011 [1 favorite]
If you saw my last company ID picture, you'd be certain of it.
All I could think when I first saw it was, "I don't really look this much like a druid priest, do I?"
posted by Kid Charlemagne at 2:43 PM on April 9, 2011 [1 favorite]
I can't speak to the biology side of things but I can confirm that *pure* mathematics is much tougher than other kinds. I studied unchaste mathematics and found it much *easier* to *get a hold of*, if you catch my drift heh heh heh.
posted by A dead Quaker at 3:16 PM on April 9, 2011
posted by A dead Quaker at 3:16 PM on April 9, 2011
Merck has a business model. It's worked in the past. If they just have the guy who turns the crank on the drug discovery box turn it a little faster everything will be fine.
Exactly. Merck's goal is to make money. They make money with drugs targeting specific proteins. It's much more efficient for them to simply test drugs to see if they work and sell them if they do, as opposed to understanding the biology in a pure way as a part of the drug design process. They say "this drug targets this protein x, having this effect on the body," but they don't really know what that drug does when it enters biology b/c its too damn complicated. Its beyond us. And for all we know, it always will be.
Why would they waste their time trying to actually understand the biology, when instead they can just create hundreds of compounds quickly, test them on rats, and see if they work? Then gradually narrow it down to a single compound with minimal side effects, but with a non-negligible effects on some "problem" you can sell people on. Then, get it through the FDA approval process and hire an ad agency to sell people on it. Make gobs of cash, success!
In fact, they would probably rather keep the black box of biology in the dark. If people actually started understanding biology better, then they might find that the drugs they are selling aren't even necessary! Or, if people could create drugs without the significant infrastructure that Merck already possesses, then they'd be out some dollars.
posted by garuda at 3:39 PM on April 9, 2011 [1 favorite]
Exactly. Merck's goal is to make money. They make money with drugs targeting specific proteins. It's much more efficient for them to simply test drugs to see if they work and sell them if they do, as opposed to understanding the biology in a pure way as a part of the drug design process. They say "this drug targets this protein x, having this effect on the body," but they don't really know what that drug does when it enters biology b/c its too damn complicated. Its beyond us. And for all we know, it always will be.
Why would they waste their time trying to actually understand the biology, when instead they can just create hundreds of compounds quickly, test them on rats, and see if they work? Then gradually narrow it down to a single compound with minimal side effects, but with a non-negligible effects on some "problem" you can sell people on. Then, get it through the FDA approval process and hire an ad agency to sell people on it. Make gobs of cash, success!
In fact, they would probably rather keep the black box of biology in the dark. If people actually started understanding biology better, then they might find that the drugs they are selling aren't even necessary! Or, if people could create drugs without the significant infrastructure that Merck already possesses, then they'd be out some dollars.
posted by garuda at 3:39 PM on April 9, 2011 [1 favorite]
The failure of all these GWAS studies is obvious and predictable once you realize they're not even paying attention to cellular circuitry. But what gets funded year after year? Another million dollars here to find low-confidence, rarely reproducible single-gene "causes."
The GWAS studies we work on look at comparing classes of things, diseases and multiple, associated genes, to look for patterns across multiple cell types. Not sure what I'm missing here, but it seems a bit of a systemic approach, to me.
posted by Blazecock Pileon at 4:27 PM on April 9, 2011 [1 favorite]
The GWAS studies we work on look at comparing classes of things, diseases and multiple, associated genes, to look for patterns across multiple cell types. Not sure what I'm missing here, but it seems a bit of a systemic approach, to me.
posted by Blazecock Pileon at 4:27 PM on April 9, 2011 [1 favorite]
Also, Eric Schadt seems like a neat scientist with some neat science, but like all scientists at that level, he's ego-tripping his balls off.
IMO, at the end of his career, we'll still be far from understanding biology as he sells it. Indeed, we will for the foreseeable, held back the non-ideality of all of the humans and human institutions that do science and the complexity of nature which has no mercy.
posted by garuda at 4:27 PM on April 9, 2011
IMO, at the end of his career, we'll still be far from understanding biology as he sells it. Indeed, we will for the foreseeable, held back the non-ideality of all of the humans and human institutions that do science and the complexity of nature which has no mercy.
posted by garuda at 4:27 PM on April 9, 2011
Jacques Monod is my hero.
That said, this is a conversation I've been having with my advisors. When I say conversation, I mean me sort of going "What happened to the big ideas? Figuring out how all this stuff works together instead of spending your time looking at the minutiae of one tiny aspect until hopefully something clicks with something else someone else is doing?"
(Can you tell I'm first year?)
I know that getting funding is all about working on projects that are very much about "small" things (small to my perhaps overblown, naive sense of importance) but when you look at what everyone was doing in The Eighth Day of Creation, I feel like the opportunity to think about this in a larger context is being buried under the need for money and publication.
I'm going to read Schadt and Ideker's papers, and this is exactly the post I needed today. So thank you all.
posted by ltracey at 4:59 PM on April 9, 2011 [1 favorite]
That said, this is a conversation I've been having with my advisors. When I say conversation, I mean me sort of going "What happened to the big ideas? Figuring out how all this stuff works together instead of spending your time looking at the minutiae of one tiny aspect until hopefully something clicks with something else someone else is doing?"
(Can you tell I'm first year?)
I know that getting funding is all about working on projects that are very much about "small" things (small to my perhaps overblown, naive sense of importance) but when you look at what everyone was doing in The Eighth Day of Creation, I feel like the opportunity to think about this in a larger context is being buried under the need for money and publication.
I'm going to read Schadt and Ideker's papers, and this is exactly the post I needed today. So thank you all.
posted by ltracey at 4:59 PM on April 9, 2011 [1 favorite]
Not sure how qualified I am to comment on this (comp science/finance background, genomics/transcription PhD, currently making a brand new version of A Popular Bioinformatics Tool), but I'll give it a stab. There's basically nothing in the article that couldn't have been written 5 or more years ago. There are hundreds of systems biologists the world over who are doing this kind of thing, and they aren't getting blowjobs from Esquire magazine.
New methods and data sources come along all the time, but the basics don't change:
* A gene list/bunch of GO terms on its own is not a result
* A few hundred transcription factors with lines drawn between them is not a result
* When the haystack is so big, probability says that more than a few pieces of straw will look very needle-like
* So verify any predictions you make as many ways as possible
We're living in a really exciting time, and there's so much interesting stuff coming out every day, but I don't think this kind of rock star science is very productive.
posted by kersplunk at 5:09 PM on April 9, 2011 [7 favorites]
New methods and data sources come along all the time, but the basics don't change:
* A gene list/bunch of GO terms on its own is not a result
* A few hundred transcription factors with lines drawn between them is not a result
* When the haystack is so big, probability says that more than a few pieces of straw will look very needle-like
* So verify any predictions you make as many ways as possible
We're living in a really exciting time, and there's so much interesting stuff coming out every day, but I don't think this kind of rock star science is very productive.
posted by kersplunk at 5:09 PM on April 9, 2011 [7 favorites]
There are hundreds of systems biologists the world over who are doing this kind of thing, and they aren't getting blowjobs from Esquire magazine.
Kersplunk, I'm pretty sure you deserve a hug from the systems biologists of the world.
posted by Slackermagee at 5:18 PM on April 9, 2011 [2 favorites]
Kersplunk, I'm pretty sure you deserve a hug from the systems biologists of the world.
posted by Slackermagee at 5:18 PM on April 9, 2011 [2 favorites]
This article about Eric Schadt is every bit a piece of vapid sales puffery as Schadt himself seems to be. Know Schadt, Sherlock! He want to bring physics to molecular biology, while at the same time disparaging reductionism. To bring physics to molecular biology he must turn instead still further to reductionism, focusing on the process of life rather than its structures. This means he must turn to the allometric scaling of metabolism, a bit of pure math he may be able to appreciate; and one that syncretizes electromagnetism and thermodynamics in a treatment of biological energy that accounts for the origins of life as well as its evolution and development. The lessons can be gleaned from the curves generated, curves representing biomass, where that biomass is an aggregation of organic molecules shaped and molded by the fluctuations in energy carried by electrical fields. In this vision the bomass is an aggregation of rechargeable batteries of organic molecules.
What Schadt cannot yet appreciate is that traditional views of bioelectricity are based upon the notion membrane voltages are due to ion concentration gradients, that is, that there is some equivalence between entropic and electrical pressure. This utter nonsense appears in the Hodgkin-Huxley Model of the nerve impulse, and the chemiosmosis of Peter Mitchell. It has been shown to be incorrect in at least three papers since Mitchell's 1978 Nobel for the idea. Mitchell was a biochemist, and knew nothing about electricity. This presumed equivalence is behind the ongoing clinical irrelevance of neuroscience which has its own biophysics, one involving complexity of structure rather than focusing on simplicity of process. Only electrical pressure and electrical fields can account for direction of microbial tail-rotor rotation, through induced magnetic fields. Entropic pressure cannot account for this, or for the generation of negative entropy in the form of battery recharge. Proton pumps acting to restore gradients just doesn't do it. Perhaps Schadt might write some groundbreaking papers on these matters. He's so heavy.
posted by Veridicality at 5:45 PM on April 9, 2011
What Schadt cannot yet appreciate is that traditional views of bioelectricity are based upon the notion membrane voltages are due to ion concentration gradients, that is, that there is some equivalence between entropic and electrical pressure. This utter nonsense appears in the Hodgkin-Huxley Model of the nerve impulse, and the chemiosmosis of Peter Mitchell. It has been shown to be incorrect in at least three papers since Mitchell's 1978 Nobel for the idea. Mitchell was a biochemist, and knew nothing about electricity. This presumed equivalence is behind the ongoing clinical irrelevance of neuroscience which has its own biophysics, one involving complexity of structure rather than focusing on simplicity of process. Only electrical pressure and electrical fields can account for direction of microbial tail-rotor rotation, through induced magnetic fields. Entropic pressure cannot account for this, or for the generation of negative entropy in the form of battery recharge. Proton pumps acting to restore gradients just doesn't do it. Perhaps Schadt might write some groundbreaking papers on these matters. He's so heavy.
posted by Veridicality at 5:45 PM on April 9, 2011
By failure of GWAS studies, I mean the failure to explain a significant amount of the measurable heritability as certain influential people had hinted at. That's not to say that there aren't some groups doing really innovative analyses, but the standard for analysis seems to be single-locus. And it's not to deny the parade of great discoveries coming out in Nature Genetics and other journals, but behind those discoveries there's a bunch of labs sitting on very expensive datasets that have little hope of getting published because they can't find any results. By ignoring the immense amount of knowledge that we have about interactions and pathways from the past decades, I think they're missing a ton of results. Here I think the blame is with the computational biologists as much as anyone else, since some pathway databases lock up knowledge in obscure formats that only the most determined can extract. If you want people to use your stuff, reduce it to a spreadsheet or table, whether it's text or Excel.
The canonical example of this is height, a hugely heritable phenotype that is the function of more time, tissues, interactions, molecular gradients, and feedback cycles than anything else we can easily measure. No one thinks that a single gene locus explains this, but that's how the analysis is performed anyway. Of course, that's just my opinion on how to go about this, and having never done GWAS personally, you should value my opinion on it as much as any other random internet keyboard warrior, but I think the GWAS field as a whole would agree that SNP chips and their analyses have not succeeded as planned.
The Emperor of Ice Cream, as a systems biologist I really do like what Sean Carroll is doing. (Though if people have problems with rah rah and Rock Star imagery, they'd really hate a Carroll talk I saw a while back.) The evolution perspective is missing from a lot of systems biology, and evolution is as powerful a tool for insight as is the cellular network. I think Carroll's outreach to non-scientists is very important, and I'm glad he's taking the time to do it. I'm not terribly familiar with Oyama and Lewontin, but prefer to stick to the data as much as possible.
posted by Llama-Lime at 5:55 PM on April 9, 2011
The canonical example of this is height, a hugely heritable phenotype that is the function of more time, tissues, interactions, molecular gradients, and feedback cycles than anything else we can easily measure. No one thinks that a single gene locus explains this, but that's how the analysis is performed anyway. Of course, that's just my opinion on how to go about this, and having never done GWAS personally, you should value my opinion on it as much as any other random internet keyboard warrior, but I think the GWAS field as a whole would agree that SNP chips and their analyses have not succeeded as planned.
The Emperor of Ice Cream, as a systems biologist I really do like what Sean Carroll is doing. (Though if people have problems with rah rah and Rock Star imagery, they'd really hate a Carroll talk I saw a while back.) The evolution perspective is missing from a lot of systems biology, and evolution is as powerful a tool for insight as is the cellular network. I think Carroll's outreach to non-scientists is very important, and I'm glad he's taking the time to do it. I'm not terribly familiar with Oyama and Lewontin, but prefer to stick to the data as much as possible.
posted by Llama-Lime at 5:55 PM on April 9, 2011
So for those who think Schadt is doing worthwhile stuff, is there a better article that presents his ideas well, for us non-science majors?
posted by kristi at 5:56 PM on April 9, 2011 [1 favorite]
posted by kristi at 5:56 PM on April 9, 2011 [1 favorite]
I disagree with him (vehemently) and I'm willing to believe that this guy is brilliant, but the article paints him as a total asshole. You know what we call grad students that date college freshmen where I'm from. Creepy, we call them really creepy.
posted by maryr at 6:09 PM on April 9, 2011 [1 favorite]
posted by maryr at 6:09 PM on April 9, 2011 [1 favorite]
...hitting 100 miles per hour on the street and 120 miles per hour at the office park and popping the occasional celebratory wheelie in nothing but the white shirt, the short pants, and the mitered black helmet that makes him look like some kind of postmodern grenadier, sporting technological plumage.
I'm sorry, I posted too soon. I feel I under-emphasized what an asshole this article makes him out to be.
posted by maryr at 6:12 PM on April 9, 2011 [1 favorite]
I'm sorry, I posted too soon. I feel I under-emphasized what an asshole this article makes him out to be.
posted by maryr at 6:12 PM on April 9, 2011 [1 favorite]
"What happened to the big ideas? Figuring out how all this stuff works together instead of spending your time looking at the minutiae of one tiny aspect until hopefully something clicks with something else someone else is doing?"
Anything involving a gene that hasn't been backed up by solid evidence of the gene's function with time consuming things like activity assays, co-localizations, DNA foot-printing analysis, and some figuring out of what regulates expression, or doesn't have an irrefutably high similarity to something that has, is bound to be bullshit. You can't figure out how the parts of an unknown system work together if you don't know what the parts are. Monod figured out the lac operon, a tiny aspect of the minutiae of E. coli fermentation and through it discovered a hell of a lot. Systems genetics only has the tools that Monod, and the hundreds that have followed him, have given it.
Really the minutiae doesn't seem so small when you consider how connected life really is and how we now have the tools to really show it. If you do end up following an adviser sweating the small stuff and you end up characterizing a gene and then putting it into gene bank you could make hundreds of genomes make that much more sense, instantly, much less the exponentially growing number that will be put into it in the future.
"Dicebat Bernardus Carnotensis nos esse quasi nanos, gigantium humeris insidentes, ut possimus plura eis et remotiora videre, non utique proprii visus acumine, aut eminentia corporis, sed quia in altum subvenimur et extollimur magnitudine gigantea"
posted by Blasdelb at 7:38 PM on April 9, 2011 [1 favorite]
Anything involving a gene that hasn't been backed up by solid evidence of the gene's function with time consuming things like activity assays, co-localizations, DNA foot-printing analysis, and some figuring out of what regulates expression, or doesn't have an irrefutably high similarity to something that has, is bound to be bullshit. You can't figure out how the parts of an unknown system work together if you don't know what the parts are. Monod figured out the lac operon, a tiny aspect of the minutiae of E. coli fermentation and through it discovered a hell of a lot. Systems genetics only has the tools that Monod, and the hundreds that have followed him, have given it.
Really the minutiae doesn't seem so small when you consider how connected life really is and how we now have the tools to really show it. If you do end up following an adviser sweating the small stuff and you end up characterizing a gene and then putting it into gene bank you could make hundreds of genomes make that much more sense, instantly, much less the exponentially growing number that will be put into it in the future.
"Dicebat Bernardus Carnotensis nos esse quasi nanos, gigantium humeris insidentes, ut possimus plura eis et remotiora videre, non utique proprii visus acumine, aut eminentia corporis, sed quia in altum subvenimur et extollimur magnitudine gigantea"
posted by Blasdelb at 7:38 PM on April 9, 2011 [1 favorite]
« Older Amo Amas Amat | Somewhere on the spectrum between disturbing and... Newer »
This thread has been archived and is closed to new comments
Also, anyone who studies the history of biology knows that there is an almost eternal, and probably never to be properly synthesized, conflict of emphasis between the geneticists (broadly defined) and the developmentalists (broadly defined): this conflict of emphasis goes back to the 19th century, and the more I read about it in all its incarnations, past and present, the more I am convinced it is simply two sides of the same story getting confused about the primary language of biological description.
posted by The Emperor of Ice Cream at 11:18 AM on April 9, 2011 [2 favorites]