Awesome Science Rant

[IainC]

-------- Original Message --------
Subject: [SyBIT-All] A farewell to bioinformatics
Date: Mon, 19 Mar 2012 10:46:06 -0700
From: Frederick Ross <madhadron@gmail.com>
Reply-To: sybit-all@sympa.systemsx.ch
To: Sybit <sybit-all@sympa.systemsx.ch>

I'm leaving bioinformatics to go work at a software company with more technically ept people and for a lot more money. This seems like an opportune time to set forth my accumulated wisdom and thoughts on bioinformatics.

My attitude towards the subject after all my work in it can probably be best summarized thus: "Fuck you, bioinformatics. Eat shit and die."

Bioinformatics is an attempt to make molecular biology relevant to reality. All the molecular biologists, devoid of skills beyond those of a laboratory technician, cried out for the mathematicians and programmers to magically extract science from their mountain of shitty results.

And so the programmers descended and built giant databases where huge numbers of shitty results could be searched quickly. They wrote algorithms to organize shitty results into trees and make pretty graphs of them, and the molecular biologists carefully avoided telling the programmers the actual quality of the results. When it became obvious to everyone involved that a class of results was worthless, such as microarray data, there was a rush of handwaving about "not really quantitative, but we can draw qualitative conclusions" followed by a hasty switch to a new technique that had not yet been proved worthless.

And the databases grew, and everyone annotated their data by searching the databases, then submitted in turn. No one seems to have pointed out that this makes your database a reflection of your database, not a
reflection of reality. Pull out an annotation in GenBank today and it's not very long odds that it's completely wrong.

Compare this with the most important result obtained by sequencing to date: Woese et al's discovery of the archaea. (Did you think I was going to say the human genome? Fuck off. That was a monument to the glory of
that god-bobbering asshole Francis Collins, not a science project.) They didn't sequence whole genomes, or even whole genes. They sequenced a small region of the 16S rRNA, and it was chosen after pilot experiments and careful thought. The conclusions didn't require giant computers, and they didn't require precise counting of the number of templates. They knew the limitations of their tools.

Then came clinical identification, done in combination with other assays, where a judicious bit of sequencing could resolve many ambiguities. Similarly, small scale sequencing has been an incredible boon to epidemiology. Indeed, its primary scientific use is in ecology. But how many molecular biologists do you know who know anything about ecology? I can count the ones I know on one hand.

And sequencing outside of ecology? Irene Pepperberg's work with Alex the parrot dwarfs the scientific contributions of all other sequencing to date put together.

This all seems an inauspicious beginning for a field. Anything so worthless should quickly shrivel up and die, right? Well, intentionally or not, bioinformatics found a way to survive: obfuscation. By making the tools unusable, by inventing file format after file format, by seeking out the most brittle techniques and the slowest languages, by not publishing their algorithms and making their results impossible to replicate, the field managed to reduce its productivity by at least 90%,probably closer to 99%. Thus the thread of failures can be stretched out from years to decades, hidden by the cloak of incompetence.

And the rhetoric! The call for computational capacity, most of which is wasted! There are only two computationally difficult problems in bioinformatics, sequence alignment and phylogenetic tree construction. Most people would spend a few minutes thinking about what was really important before feeding data to an NP complete algorithm. I ran a full set of alignments last night using the exact algorithms, not heuristic approximations, in a virtual machine on my underpowered laptop yesterday afternoon, so we're not talking about truly hard problems. But no, the software is written to be inefficient, to use memory poorly, and the cry goes up for bigger, faster machines! When the machines are procured, even larger hunks of data are indiscriminately shoved through black box implementations of algorithms in hopes that meaning will emerge on the far side. It never does, but maybe with a bigger machine...

Fortunately for you, no one takes me seriously. The funding of molecular biology and bioinformatics is safe, protected by a wall of inbreeding, pointless jargon, and lies. So you all can rot in your computational shit heap. I'm gone.

--
Frederick Ross
http://madhadron.com/

[Bunk]

Would have had more impact if he hadn't made up the word "ept" in his first sentence.

[Murgos]

Would have had more impact if he hadn't made up the word "ept" in his first sentence.

I actually paused reading at that point and went a googled ept just to make sure it wasn't really a word.

I did enjoy the rest of the rant though.

[Sky]

So, an IT person who thought he was better than those he served. I'm shocked.

[ghost]

So, an IT person who thought he was better than those he served. I'm shocked.

Maybe we should send him an f13 invite.....

[RhyssaFireheart]

Would have had more impact if he hadn't made up the word "ept" in his first sentence.

I actually paused reading at that point and went a googled ept just to make sure it wasn't really a word.

I did enjoy the rest of the rant though.

I think he meant "apt" which is a word.

The rest of it, yeah, "IT person thinks he's better than those he supports" is about right.

[Lantyssa]

Bioinformatics really is mostly shit, though.

[TheWalrus]

I'm pretty sure he meant ept, as in calling the department he worked for, inept. It's subtle I know.

[Merusk]

I think he meant adept and simply failed at hitting two keys.  "...a software company with more technically ADept people and for a lot..." makes more sense than the previously suggested corrections.

[ghost]

It is perfectly reasonable to think that the author thought "ept" is a word.  Does it matter?  It's still a beautiful rant.

[01101010]

It is perfectly reasonable to think that the author thought "ept" is a word.  Does it matter?  It's still a beautiful rant.

Someone slay this heretic! 

[Sky]

Rants are so 1999.

[ghost]

Rants are so 1999.

Too bad WUA left.   

[pxib]

Yeah, this is a pretty standard IT rant. It's still hard to overstate how monumentally disappointing gene sequencing has turned out to be. Rather than opening up a Brave New World of genetic engineering, it basically just revealed the staggering dimensions of our ignorance about fundamental biochemistry. We thought that reading the genetic code would be the answers in the back of the book, and it turned out to be another book entirely in a language nobody has ever seen. Pouring people at the problem (and all sorts of excited young scientists jumped at the chance to get in on the ground floor of the scientific revolution that never came) hasn't produced a Rosetta Stone, it's just created mountains of data...

...that we can't read well enough to confirm whether it's even spelled right, much less whether it's crucial code or "junk DNA".

We don't even know what junk DNA is.

The idea that piling enough meaningless data together will eventually produce a visible pattern assumes that our genetics are just cyphertext... messages purposefully designed by humans minds to be turned into readable language by relatively simple algorithms. Everything we've found out in the last decade makes it very obvious that DNA's chemistry is nothing of the sort.

[apocrypha]

Actually there were plenty of people who said a long, long time ago that it was a waste of time. At the start of the Human Genome Project many scientists were adamant that the only purpose of HGP was to make vast sums of money for the manufacturers of sequencers, and that functional analysis was a far better route to take. They were dismissed as conspiracy theorists and naysayers and ignored.

That guy is right, bioinformatics is a ridiculous discipline that only exists to support the absurd biological reductionism behind mass sequencing, most of which is useless. But hey, it makes biotechnology companies really, really rich.

[Kitsune]

Given the sheer number of roles that DNA performs in cells, it's like a blueprint made out of power tools.  Calling any of it 'junk' at this point serves more to highlight our ignorance than any likely uselessness that exists in it; it's perfectly likely that the junk parts are still doing something important that we just don't yet understand.  I'm reminded of how Egyptians threw out brains when embalming mummies, considering them useless organs.  We've come a long way in biosciences in the last few decades, but there's a whole lot longer to go yet.

[Sky]

It's still hard to overstate how monumentally disappointing gene sequencing has turned out to be. Rather than opening up a Brave New World of genetic engineering, it basically just revealed the staggering dimensions of our ignorance about fundamental biochemistry.

Revealing ignorance is the first step to enlightenment. Rather than disappointing, it's the beginning of that school of science as much as a shaman dishing out herbs to tribe members was or a sawbones on a civil war battlefield was to surgery.

Discoveries don't work on a schedule and you can't rush science.

[ghost]

It's funny to think that people had the audacity to call it a waste of time.  Every amazing discovery starts with a small bit of information.  I think the most of the disappointment regarding the project stems from the completely unreasonable expectations that were tagged to it.  I suppose that stems, in a large part, from how funding is currently obtained for said projects.  It wouldn't have gone to completion had they said, "hey, we may get a lot of information out of this project that we won't understand at all and may not be able to use for many, many years," because nobody would have funded that shit.

[Lantyssa]

Studying genetics is not a waste of time.  I've known a lot of people who have done some very interesting work with it.

However, that are a lot of people studying genetics that are a waste of time.  The sequencers are mostly a boondoggle, too.  Only a few groups really use them to do good things.

[K9]

I'm a Bioinformatician and all I see is some butthurt IT nobber having a rant because biological reality doesn't mesh neatly with his OCD view of how systems should work. He can fuck off to his shitty startup making Angry Pigs for Android and enjoy it.

That guy is right, bioinformatics is a ridiculous discipline that only exists to support the absurd biological reductionism behind mass sequencing, most of which is useless. But hey, it makes biotechnology companies really, really rich.

Do you have ANYTHING to back this up? Do you have any experience in the field or any technical knowledge which could in any way justify this completely ridiculous assertion?

Yeah, this is a pretty standard IT rant. It's still hard to overstate how monumentally disappointing gene sequencing has turned out to be. Rather than opening up a Brave New World of genetic engineering, it basically just revealed the staggering dimensions of our ignorance about fundamental biochemistry. We thought that reading the genetic code would be the answers in the back of the book, and it turned out to be another book entirely in a language nobody has ever seen. Pouring people at the problem (and all sorts of excited young scientists jumped at the chance to get in on the ground floor of the scientific revolution that never came) hasn't produced a Rosetta Stone, it's just created mountains of data...

...that we can't read well enough to confirm whether it's even spelled right, much less whether it's crucial code or "junk DNA".

We don't even know what junk DNA is.

The idea that piling enough meaningless data together will eventually produce a visible pattern assumes that our genetics are just cyphertext... messages purposefully designed by humans minds to be turned into readable language by relatively simple algorithms. Everything we've found out in the last decade makes it very obvious that DNA's chemistry is nothing of the sort.

The term Junk DNA has been defunct for several years now, and I wouldn't say that the progress of molecular genetics and bioinformatics has been disappointing. It's just that the initial assumptions about how tractable biological systems are have proven to be grossly oversimplified. That's not a failure of science, it's just an example of how wondrously and infuriatingly complex nature is. I'm not surprised that some database-monkey got all bent out of shape because life won't fit into his neat little boxes.

Studying genetics is not a waste of time.  I've known a lot of people who have done some very interesting work with it.

However, that are a lot of people studying genetics that are a waste of time.  The sequencers are mostly a boondoggle, too.  Only a few groups really use them to do good things.

There's a lot of people doing anything that are a waste of space, that's humans for you. The sequencing bloat isn't ideal, but then we're stuck in a field where first principles mean fuckall, and most attempts through history to provide a first-principles approach to genetics and such have proven to be naive and/or bollocks. So we search the hard way, and learn inefficiently. It's not always elegant, but that's because biology isn't maths, it's not ordered and elegant.

/rant

[pxib]

I wouldn't say that the progress of molecular genetics and bioinformatics has been disappointing. It's just that the initial assumptions about how tractable biological systems are have proven to be grossly oversimplified. That's not a failure of science, it's just an example of how wondrously and infuriatingly complex nature is.

Well sure. Great unknowns remain as thrilling to scientists as the empty spaces on maps are to explorers. It's not the scientists who are disappointed... it's their funders and the general public. An enormous amount of private and government money poured into gene analysis around the turn of the century, and an enormous amount of talent followed it into research and business.

No revolution followed -- no great big Moon landing event we could all crow about -- so a lot of people are wondering what happened to all the cash. It's the standard "Yes, they're important, but nobody actually likes negative results" problem that sponsor-based science always has... but writ large by the proximity and the scale of undelivered promise.

[Sheepherder]

Given the sheer number of roles that DNA performs in cells, it's like a blueprint made out of power tools.  Calling any of it 'junk' at this point serves more to highlight our ignorance than any likely uselessness that exists in it; it's perfectly likely that the junk parts are still doing something important that we just don't yet understand.  I'm reminded of how Egyptians threw out brains when embalming mummies, considering them useless organs.  We've come a long way in biosciences in the last few decades, but there's a whole lot longer to go yet.

Appendix.  Wisdom teeth.  Ear muscles.  Tailbone.

Plus they've tested some of this hypothetically useless genetic information on species that can't beg for the sweet release of death.

[apocrypha]

That guy is right, bioinformatics is a ridiculous discipline that only exists to support the absurd biological reductionism behind mass sequencing, most of which is useless. But hey, it makes biotechnology companies really, really rich.

Do you have ANYTHING to back this up? Do you have any experience in the field or any technical knowledge which could in any way justify this completely ridiculous assertion?

I was  a molecular biologist for 15 years and saw research grant after research grant after research grant poured into blanket sequencing projects, blind microarray assay projects and massive genome labs that actually produced pretty much nothing of value in that entire time.

I studied at Leeds Uni Genetics department and pretty much then entire staff there felt that their discipline was being gutted to make way for endless sequencers. The obsession with sequencing was splitting the geneticists and the molecular biologists apart from all the other disciplines and leaving them isolated. This continued in every department I ended up working in after leaving Uni. I never worked in a pure molecular biology role, but always on clinically orientated research projects - reproduction, vascular medicine, microbial antibiotic resistance, tissue engineering, and the job always, always ended up with non-molecular biologists insisting that just sequencing everything was the thing to do.

What this meant was that huge chunks of research money were just being spent on outside sequencing labs. The hospitals & universities I worked in never had enough money to have their own high-throughput sequencing systems, so we paid for commercial labs to do it. I ended up just being a liaison between clinical professors and industrial bioinformaticists, watching public money being thrown at sequencing machines in outside companies. I found it demoralising and pointless. It meant that having a discussion about the actual factors that cause the problems with whatever illness/condition/etc we were supposedly studying were nigh on impossible.

The result is biological reductionism and it stifles understanding. IMO.

[K9]

Ok, I'll wind back my complaint then, sorry. I still maintain that the OP is a massive tool though. I guess your experience doesn't match with mine. I disagree that the drive to sequence everything has stifled understanding. The challenge in genetics is that pretty much all phenotypes are the sum of a large number of weak effects that you cannot decompose solely through analyses of external conditions. You have to sequence crazy amounts of stuff to begin to see any trends. Perhaps the difference is that I (being a little younger than you) was never sold the idea that genetic sequencing would lead us to a nirvana of human perfection. To me, and in my experience, sequencing has been a valuable tool that when applied well (and it often is) allows us to dig into the most complex problem humanity has available.

I work on the evolution of pathogens, and here I see the awesome power that large quantities of sequence data have to enable us to reconstruct the dynamics of almost all pathogens. Without sequence data it would have been nigh-impossible to work out the origin and transmission rate of A/H1N1 within a week or two of the start of the pandemic. Without sequence data we wouldn't have been able to reconstruct the source of the recent EHEC outbreaks in Germany and France, nor pinpoint the source of all the recent Cholera epidemics to water from the Bay of Bengal. We wouldn't be able to understand why some vaccines work, and others fail; how the effects of some vaccines drive drug resistance and virulence and others work as intended. Without sequencing we could not be looking at the gut microbiome of humans, and starting to understand the significance of the interaction between the gut flora and human health. The advent of the MiSeq and MinION platforms we are looking at a future where instant diagnosis of infection is possible, minimising the chances of incorrect therapies and driving down the effects of drug resistance as well as reducing time spent in hospital. A future where epidemics and pandemics can be monitored and controlled in real time.

Sure there is a lot of pie-in-the-sky ideas surrounding sequencing, and sure there is a lot of pointless and wasteful work done. However there is a core problem in genetics that it is impossible to know a-priori what the size of a sufficient data set is; so you gather data until trends emerge. It is unfortunate that most traits are far more cryptic than people ever imagined, but that does not mean that the work has no value.

[apocrypha]

Oh I totally agree that the ranting guy is a bit of a tool, not entirely sure what he thought he was going to achieve other than letting some steam off.

And yes, you're absolutely right, applied well and done sensibly sequencing is vital and we'd have a far poorer understanding and set of tools available to us now without it. I was overly harsh in my condemnation of the entire field of bioinformatics, sorry, I'm kinda jaded!

I suppose my main issue is, as you say, the way it was sold - particularly in the early 90's when the HGP was gearing up. While many in the actual field saw through the transparent advertising of the medical/industrial technology sector a lot of people in other areas bought the sell. Particularly people like senior clinicians, politicians, university chancellors, etc. People who controlled the purse strings! An over-reliance on one tool in isolation is never a good thing and it felt like that was what happened on a large scale.

[ghost]

It all comes back to the fact that sequencing was oversold as a technology.  I'm sure that eventually we'll understand what all of the information that we obtained from the research means, but we were expecting it to explain everything and clearly it doesn't.  There are too many gaps.  For instance, there are genes that encode for proteins in bones, but there is no "femur" gene or "mandible" gene.  How do the cells know to make a mandible if it isn't coded in the DNA.  There are so many things that happen in development that we just don't understand fully.  Alternative RNA splicing seems critical for a lot of processes in development.  I'm sure there are other things going on that we have no idea about at this juncture.  I can, however, see how it would be easy to get jaded on sequencing if you were working on another important line of research that lost funding or space to make room for a sequencing lab.  That doesn't change the fact that a lot of really, really cool shit has been done with sequencing and that the information undoubtedly will become more and more useful as time goes on.

But really this is all just a waste of time.  I know the answer:  Midichlorians.   

[Lantyssa]

There's a lot of people doing anything that are a waste of space, that's humans for you. The sequencing bloat isn't ideal, but then we're stuck in a field where first principles mean fuckall, and most attempts through history to provide a first-principles approach to genetics and such have proven to be naive and/or bollocks. So we search the hard way, and learn inefficiently. It's not always elegant, but that's because biology isn't maths, it's not ordered and elegant.

It doesn't help that if your project fails to include some biology connection, your chances for funding go way, way, way down.  It's really fucked with chemistry grants over the last decade, and it's not doing the biological sciences any favors either.

And while it might be very complicated, I think saying it's not ordered and elegant is doing the field a disfavor.  People are just hesitant to throw out a decade of data and start over with a more methodical approach.  (Not that I blame them, I still have my notebooks from twenty years ago.)

[01101010]

It doesn't help that if your project fails to include some biology connection, your chances for funding go way, way, way down.  It's really fucked with chemistry grants over the last decade, and it's not doing the biological sciences any favors either.

Not to get too far out of line here, but this is so very true. Every psychiatrist I have associated with/worked for is now scrambling to include some sort of biological markers to tie to their research grants, whether it is brain scans, blood draws, or saliva collections. Seems biology is seeping into everything... which, in terms of psychology, means more intrusion by big pharma. Behavior is messy, but if you can forge a biological link, even on the shakiest of data, you will have pill makers drooling over you.

[ghost]

Applying for research grants has become such a scam.  It seems like in the old days people researched things they were actually interested in.  Now they just follow the money. 

[Nebu]

Applying for research grants has become such a scam.  It seems like in the old days people researched things they were actually interested in.  Now they just follow the money.  

It's a scam in more ways than you know.  The good-ol-boy system is alive and well, particularly when it comes to research funding.  You have to be doing the right science (ie trendy) at the right school, with the right pedigree, and know the right people at NIH/NSF if you want to maintain any kind of research program.  

As for bioinformatics: It's a necessary field and provides useful information when done by an expert.  Problem is: We have about 1000x too many of them out there as the software is VERY low level and easy to use and schools like having one on staff as the research is cheap to conduct.

[01101010]

Applying for research grants has become such a scam.  It seems like in the old days people researched things they were actually interested in.  Now they just follow the money.  

Well duh. But the other part of it is, funding is channeled to certain research. Trying to get funding for what you are interested in when the trends toward your topic are dead or ignored is stupid. Grant apps get tossed for the dumbest of reasons, including not having a certain author's research quoted somewhere, not having a certain methodology outlined, not a topic of interest. I tried my damnedest to do my graduate work on medical marijuana and the changing social/political climate on the topic back in 99 when it was not hot topic point. I was "discouraged" in following it any farther than a few papers for classes. And when I raised the issue as a valid topic for my thesis work, it was shot down extremely quickly. Of course if I had waited 10 years, I would probably be working in that field today... trying to add a biological element to my grant proposals.  

edit: damn Nebu... beat me to it. 

[pxib]

In the old days biology research required a notepad, a microscope, and an Erlynmyer flask. Physics research could be done in the back yard. Chemistry reagents were available at the local pharmacist. Those great unknowns were everywhere, and relatively easily accessible to any student of inquiry who went inquiring. Today the gaps are tiny and even perceiving them requires either precision equipment or massive sample sizes... or both.

That requires a lot of money.

Functionally science is in the same place as the movie industry or AAA game studios: The risk part of Risk/Reward has gotten so expensive that unless there's some built in reward (being a sequel to a popular franchise, say) it's almost impossible to get funded. If you want to go your own way and discover a new species of leech in the puddles in your back yard, you could probably write a grant for a hundred dollars. Heck, you could probably fund yourself. If you want to look for the Higgs boson in the locally strong gravitational field left for a few fractions of a second after a relativistic collision... you're probably not building your own gear in your garage.

Scientists didn't get greedy and funding organizations didn't get stingy. The cutting edge just keeps getting harder and harder to see.

[Khaldun]

I like the speed with which you guys got to the reasonable middle, which is the right place to be here. I do agree especially that the original ranter is one of those "Waaaaaat? Life and the universe doesn't have neat laws, simple foundational principles, and a reality that is tractable to clear and precise quantification? Fuck that shit." kind of characters. I was giving a talk recently on the impact of complexity theory, emergence and so on in social science recently and there were definitely a few folks in the audience who were profoundly freaked out by the idea that there might never be a simple way to reliably create or ensure desired outcomes in a genuinely emergent phenomenon or circumstance--basically the same people who look at a complex social situation and want to run regressions until they find the single "real" causal variable, so that they can then turn around and sell that back to think-tank policy wonks the way that snake oil salesmen used to sell quack medicine to their marks.

[ghost]

In the old days biology research required a notepad, a microscope, and an Erlynmyer flask.

And that's the way it should be, goddammit!  

What I found interesting, during my last go round of forced research internment, was the amount of things that I would consider typical lab fare for peons like myself that was farmed out to outside labs.  It made things much more tolerable than when I was doing a lot of that crap myself.  It shows how much things have changed when it's cheaper to outsource than to have a lab monkey do it.  

[Kitsune]

For instance, there are genes that encode for proteins in bones, but there is no "femur" gene or "mandible" gene.  How do the cells know to make a mandible if it isn't coded in the DNA. 

My in-no-way-a-biology-professional opinion is that we're fractals.  That we'll discover that DNA is holding an equation that works out to be a human when taken to a sufficiently high iteration rather than a specific blueprint of every nook and cranny of a body.