Skip to main content
Menu

Biological News

Farewell to Bioinformatics

Here are some angry views that I don’t necessarily endorse, but I can’t say that they’re completely wrong, either. A programmer bids an angry farewell to the bioinformatics world:

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.

That’s unfair to molecular biologists, but is it unfair to the state of bioinformatic databases? Comments welcome. . .
Update: more comments on this at Ycombinator.

62 comments on “Farewell to Bioinformatics”

  1. anon the II says:

    I think it’s probably fair to be highly skeptical of any computational method where a bunch of data goes in and answers come out. If you don’t understand the algorithms, you have no idea what’s going on. I remember learning that the Tripos force field didn’t contain a polynomial to describe torsion terms. No wonder a gauche butane interaction was about 10-fold off. However, I watched for the next 15 or so years as scientist described results using the garbage that was the Tripos force field.
    This stuff goes on all the time. I think it’s the primary reason that a lot of fairly liberal scientist don’t completely buy into the global warming stuff. They suspect that it may be true and they like the idea of lowered pollution, sustainability, etc. But they don’t have any faith in the computational models because they’ve seen how flawed and deceptive they can be with systems simpler than the weather. And why would a scientific genius go into meteorology?

  2. Teddy Z says:

    I don’t agree with #1 in regards to his comments on Global Warming (that seems like a WTF non-sequitur). However, in terms of the biologists not understanding the algorithms, he is right on. Bioinformatics is a blackbox and I think most biologists missed their introductory compsci class when they taught GIGO. I wrote about something similar here: http://www.quantumtessera.com/know-your-screen/

  3. Calvin says:

    Oh that did make me laugh. I can almost imagine the vitriol that he is directing towards all those shitty results.
    I remember my early days in antibacterials where we poured over bioinformatics on essential genes. Except it turned out they weren’t so essential. This was waved away with “Oh but they’re conditionally essential”. They were essential if you assume the input data wasn’t utter guff. Ah, those were the days.
    #1 Yup. You described my views on Global Warming. Perfectly.

  4. jane says:

    As a fairly liberal scientist, I will just say that I buy into the global warming stuff because we never used to have 70-degree weather and tornadoes in January in my neck of the woods when I was a kid, whereas now it’s becoming a regular occurrence.
    The quote mentions GenBank. I have heard some strong complaints about GenBank by molecular-biologist colleagues recently. Many sequences do not have voucher information attached, so there is no way of independently confirming their identity, and some are certainly mislabeled. Makes you wonder about some of this barcoding.

  5. Morgan Price says:

    What a pessimist. There are a lot of errors in gene annotation databases but they are still very useful. It would be a lot of work to annotate every gene I come across myself! And even an incorrect annotation can be a useful hint. Similarly, gene expression data is not as useful as first hoped, but you’d be foolish to study gene regulation without taking advantage of it. Also, recent data (from high density oligo arrays or from sequencing) is more reliable and quantitative.

  6. I think the criticism’s partially fair; a lot of computational biology is big on fancy, abstract mathematical models and short on predictions that can be tested at the molecular level to design new therapies. Plus there’s also the occasional annoying math whiz who thinks he or she can revolutionize drug discovery using his latest network model.
    That being said, the field is still new and is struggling to find its optimum benchmarks and areas of application, so I think the criticism is a little unfair. The hope is that twenty years down the line something real and robust would emerge from the networks and algorithms. My guess is that just like it happened in computational chemistry, computational biologists will identify specific, niche areas where the models work. The main criticism of the field comes from people who tend to overestimate its importance. We just need to understand that these are all tools and nothing more.
    Also, @1, “And why would a scientific genius go into meteorology?”
    John von Neumann – a bonafide genius by pretty much any definition – was one of the pioneers of meteorological modeling.

  7. Anonymous says:

    What the hell is a ‘Liberal Scientist’? Either you are a Scientist or Not! Any gray are just makes you a joke.

  8. SP says:

    I view gene annotations the same way I view primary HTS data. You don’t want to rely heavily on any one result, but overall the data is likely pointing in the right direction.

  9. John Wayne says:

    My issues with the global warming debate are as follows:
    1. We haven’t been able to pollute another earth-like planet to see what happens, so any models put forth are going to speculative at best. These models aren’t necessarily wrong, but we really have no idea.
    2. The issue has become about politics and money, so the issue has become polarized and clouded with BS.
    And most importantly, 3: The global warming debate has somehow overshadowed the facts that burning oil pollutes our environment to our direct detriment, we are going to run out of oil in the short term, and oil is too valuable as a feedstock to burn. We need a source of clean energy ASAP, and it doesn’t matter at all if global warming is real.

  10. Brian says:

    Gotta link the very lively Ycombinator thread on this post:
    http://news.ycombinator.com/item?id=5123022

  11. Your excerpting is quite generous to the author; you’ve left out some of his most ridiculous bile.
    Yes, there are a lot of problems in bioinformatics, and yes there are a lot of slow learners. There’s a lot of rotten software, and a bunch of land mines that each generation re-learns (such as having gene names such as Oct9 converted to dates by Excel).
    But, to throw the baby out with the bathwater, as many of the portions you left out do, is absurd. Bioinformatics DOES make real discoveries. DNA sequencing really CAN be used to advance science. Good scientists in the field really DO know about these pitfalls, and really DO work around them.
    Every field has a lot of weak practitioners of poor science, and they all get it published. To write off an entire field by looking at the worst of the muck done by the most careless, rather than a balanced analysis of what works and what remains to be done, is a waste of bytes (and certainly of the readers’ time).

  12. dearieme says:

    “DNA sequencing really CAN be used to advance science.”
    Excellent.
    Can it be used to cure many ill people yet?

  13. Perdurabo says:

    Every field and theory will at some time be asked stand up and answer for the results it has or has not shown. If there remains a dearth of valid results but the underlying theory is still sound, perhaps the academicians will in the future come up with an application showing better practicality, but for now maybe it would be best to not use it to direct Big Pharma to find new drugs.
    The quoted passage reminded me of a similar one posted in CEN years ago in regards to quantum mechanics. The writer said it was time to “send it off with an appropriate whack on it bedraggled backside”, or something to that effect

  14. Anonymous says:

    John Wayne: Agreed, views on global warming have become clouded with bullshit. Let’s take a look at the literature to see if that helps.
    Of the 13,950 peer reviewed papers that mention climate change published between 1991 and 2012:
    0.17% reject global warming.

  15. Am I Lloyd peptide says:

    Not to make this comments section about climate change, but to those here who are knocking on global warming computer models, you need to realize that there’s a vast amount of independent fieldwork that also confirms (or sometimes refutes) what the models say. Climate change skeptics (even reasonable ones) often give the impression that most of the conclusions about climate change stem from computer models, but that’s just not true; for every armchair modeler there’s five scientists out there in the Arctic and in the rainforests, looking at species and ice loss. It’s very much an experimental science.

  16. pete says:

    @12 dearie
    – s’best not to talk on an empty head, laddie

  17. Twelve says:

    As a conservative who is a scientist, I also acknowledge global warming as an established fact, and I see anthropogenic CO2 generation as the likely main cause. I’m not a climate scientist, so I have to take the word from the 99% or so of competent climate scientists who say that human activity is driving these changes.
    It’s true that I can’t vouch for the algorithms that they use. I can’t do that for the ones that allow my car to run either, but given the quality of the observables (in both cases), I am OK with that. Impugning the validity of climate scientists’ methodology without a coherent criticism is akin to mudslinging in politics – it reflects only on the perpetuator, not the target.

  18. pgwu says:

    Working in biophysics and in drug discovery since 1990’s, I have seen garbage-in-garbage-out a few times, both on the lab side and on the computer side. With increasing computing power and availability of tools, some people have made this into an art form. It has nothing to do with a person’s occupation. I like the idea proposed in the link: http://biostatistics.oxfordjournals.org/content/10/3/405.full

  19. road says:

    @15 Am I Lloyd peptide
    – That’s a fair point, but I think what some people are skeptical of is the projections about future changes (rising sea-levels, etc). Reasonable people don’t doubt the empirical facts that things have been changing lately (increased temperature, CO2, etc), but I’m not convinced that anyone knows with any certainty (1) that it’s anthropogenic, or (2) what will happen decades from now. And the only way to address those questions is with modeling.

  20. Ed says:

    #14 if you want to convince skeptics that global warming is real, probably best to link to “evidence” from somewhere other than treehugger.com, and make sure that it is written by someone with a degree in a scientific subject (unlike Michael Richard that authored that article you linked to – degree in Law)

  21. John Wayne says:

    @14: I agree with your conclusion, but not how you got there. I do not agree that the number of papers published in a political field wherein funding is at stake is predictive of the truth.
    We have a huge energy problem that must be addressed immediately, and we’re wasting time having a debate (global warming) that doesn’t matter.*
    *caveat: global warming outside of CO2 production from fossil fuel is still a thing. My philosophy is that we can’t really know, but it is better to not risk the earth when we can try do things a different way that has other direct benefits.

  22. DannoH says:

    Isn’t the real root cause of global warming the sun? Can’t we just install a rheostat on its output, or put a big sunshade up, or perhaps some kind of modified tox’u’tat?
    End Sarcasm =)

  23. Arthur Dent says:

    The reference to 0.17% of papers disagree with the CAGW theory is hugely misleading. As is most of the hot air (:-)) expended in the debate on partisan blogs. There is scientific disagreement but also a large degree of agreement as well.
    The vast majority of scientists on both sides of the debate agree with the following|:
    · that climate has always changed and always will,
    · that carbon dioxide is a greenhouse gas and warms the lower atmosphere,
    · that human emissions are accumulating in the atmosphere,
    · that a global warming of around 0.5OC occurred in the 20th century, but
    · that global warming has ceased over the last 15 years.
    The scientific argument over CAGW is therefore about none of these things. Rather, it is almost entirely about three other, albeit related, issues. They are:
    · the amount of net warming that is, or will be, produced by human-related emissions,
    · whether any actual evidence exists for dangerous warming of human causation over the last 50 years, and
    · whether the IPCC’s computer models can provide accurate climate predictions 100 years into the future.
    Quote from Professor Robert Carter

  24. Anonymous says:

    This is not a new finding! Garbage in – Garbage out is a well-known philosophy in modeling world. I assume this person knows this before he entered the bioinformatics field.

  25. Cooter says:

    Former big pharma head of bioinformatics here. There are more than a few bits of truth in Ross’ rant, but none of them are new: people in the field have been saying these things in one form or another since the 90s, at least. In my own experience, the greatest excesses in bioinformatics have come about because politicians and senior executives (sometimes the same people, and usually not biologists) hear that something in bioinformatics is the Next Big Thing (TM), and We Will Lose our Competitiveness If We Don’t Invest Now (TM), and then they fund huge initiatives that fly in the face of common sense.
    Anyone hear things like, “Big Data is the Next Big Thing!” and “We Need to Invest in Data Scientists!” lately? Same damned thing.
    At the same time, I have a lot of experience with physicists who can’t get a job in physics, so they become scientific programmers and bioinformaticists. The physicists often seem to “know” exactly how to “solve” the molecular biologists’ problems through the use of large datasets and complex statistical approaches, while the molecular biologist is just asking for an easy-to-use workflow tool to help them plan their restriction digests.
    I’ve found that physicists’ prescriptions for biologists often have a lot in common with Andy Grove’s prescriptions for the pharmaceuticals industry.

  26. pgwu says:

    The discussion here resembles what happen in some drug discovery/developement project meetings.

  27. JRnonchemist says:

    Re: Global warming
    The phrase ‘global warming’ implicitly sets the system being measured as the whole earth. Think about the geometry, and to how much an average temperature of the whole earth would be influenced by the hotter bits towards the center.
    My understanding is that our ability to measure the interior temperatures is such that it has relatively large inherent measurement error.
    In contrast, we can define a volume as ‘The Earth’s Surface’ such that we can hope to be able to measure it with a usefully low inherent error.
    This is an important and significant distinction.

  28. Nekekami says:

    The problem with bioinformatics is two-fold: One, as #24 said: Garbage In Garbage Out.
    The other is a mix of issues into one compound problem: Problem complexity and overly aggressive attempts to isolate and abstract, thus discarding a lot of potential connections between data, which further reinforces the GIGO problem.
    As for the IPCC models, I’m under a court order preventing me from disclosing a lot of details, but I was part of an investigation initiated by Forskningsrådet(the Swedish national research council) into scientific misconduct by SMHI(Swedish meterological institute) and their climate research. It was shut down by the Justice Department before even finishing, and as a result the codebase is now only available to personell with clearance from the climate research department at SMHI. Some of the algorithms were considered to be inaccurate and obsolete already in the 70’s. Some algorithms, when analysed as FSM’s, failed to resolve properly, and it turns out that the libraries and compiler relied on “magic numbers” to yield not-completely-insane returns.
    One dangerous fallout of SMHI going all-in with IPCC is that they now rely on fewer weather stations for WEATHER predictions, not just climate predictions, reducing the quality of them. As such, the Swedish Airforce and Luftfartsverket(civil aviation authority) have set up their own network of weather stations, with over 200 stations, instead of SMHI’s less than 20). The Swedish Navy have dropped most of their collaboration with SMHI and gone almost completely in-house too, because SMHI became too unreliable, and instead work together with many civil shipping companies to produce more detailed weather forecasts.

  29. Jonathan says:

    @12, yes, what a ridiculous comment to make. BRCA testing and hereditary breast cancer, Xalkori and CF patients with the G551D mutation, plenty of examples in the rare disease world that are growing every day (but we can use Noah and Alexis Beery, and Nik Volker as the literal poster children), AIDS patients taking abacavir, the use of DNA sequencing in stemming the deaths at the NIH clinical center after the KPC outbreak…
    None of those are possible without sequencing. But I suspect you knew that already.

  30. gippgig says:

    Murphy’s Law of Databases: Databases are riddled with errors. It isn’t just bioinformatic databases.
    Bioinformatics has limited value for drug development but is very valuable for scientific discovery. Been there, done that (I was doing iterative consensus searching by hand long before it was built into programs).
    As for global warming, the choices are to (a) use the models as the best guesses we have or (b) stick your head in the sand and hope everything turns out OK. At the moment it looks like the models are indeed wrong – the actual situation is worse than predicted.

  31. anon the II says:

    I apologize for mentioning global warming. Please forgive me. I was just using it as an example where complex algorithms are used that I don’t really have the time or expertise to evaluate and understand.

  32. Oldnuke says:

    Reminds me a lot of the “work” performed at the National Laboratories on the nuclear stockpile.
    Everyone seems to have forgotten (and maybe most of us have died off by now) just how BAD computer models were in comparison to the actual results of nuclear testing.
    Of course, now there is (for all intents and purposes) NO real testing and the models are just grand.
    All we’ll need is the next generation of Cyber-Phallic Symbol (CPS) to make the next leap forward.

  33. ajnorton says:

    “Bioinformatics has limited value for drug development but is very valuable for scientific discovery”
    I imagine it’s useful when assembling the monoclonal antibodies that seem to be doing better than most small-molecule stuff.

  34. Shanedorf says:

    There are many here with a lifetime of scientific training, so I really wanted to ask this simple question
    If we accept that “for every action, there is an equal and opposite reaction”…then what exactly is the equal and opposite reaction of combusting billions of barrels of previously inert (underground) hydrocarbons into the closed system known as our atmosphere ?

  35. drug_hunter says:

    two words: “hairballomics”
    Bioinformatics will be incredibly useful someday (when we have more/better data) but, sadly, it seldom is more than a very blunt instrument at the moment.

  36. newton says:

    34. Shandof:
    “If we accept that “for every action, there is an equal and opposite reaction”.”
    That’s an archaic (actually, the original) formulation of the law of conservation of momentum. It is not some over-arching principle of nature that one can apply in some philosophical way to the environmental effects of burning oil. So, no, we don’t accept it in that sense.

  37. MIMD says:

    Re: 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.
    Sorry to see this.
    A number of years ago, I wrote the piece “Has Bioinformatics Hit A Hard Wall of Stagnation?
    I’d also penned the Bio-IT WOrld letter seen in that post, “Medical Informatics MIA”, while at Merck Research Labs, calling for more clinician-bioinformatician collaboration.
    Sorry to see this current letter.

  38. JRnonchemist says:

    @23 Arthur, can you tell me more about this Professor Robert Carter?
    @28 US or Swedish Justice Department?
    @31 If you want to talk about a problem involving mathematical modeling, statistics, and measurement error, the subject you mention is sure to have name recognition. It is also relevant to the political situation in at least some countries, and has expected consequences from that.
    On one side you have people saying ‘you are crackpots who don’t understand thermodynamics, and hate poor people’. On the other side, you have people saying ‘you are crackpots who don’t understand thermodynamics and hate poor people’.
    @34First, it isn’t a closed system. There is matter addition from space and from volcanic and otherwise outgassing. There may be some small amount of matter loss to space. (Light, hot stuff in the upper atmosphere.) Then there is energy transfer. I’d guess density is a bit low for tidal forces to do much work directly, but there should be a heat flux from the earth. And a bunch of other stuff besides, depending on where you set your boundaries.
    Secondly, calling fossil fuels inert has some issues. That carbon was in the atmosphere in the first place. It got buried under the ground long ago, and has been escaping ever since. The stuff that is still there is just the stuff that remains after having been escaping for millions of years. I certainly couldn’t say how much of the stuff entered the atmosphere before human civilization, or what all the effects of it were.
    So, a) I’d suggest just finding a good thermodynamics book, maybe suited for an upper division engineering course, and solving problems in it. b) a good heat transfer book might help with understanding the time dependence aspects. Failure to specify time frame leaves it a variable, which means that for the sake of argument, extreme cases can be used.
    Maybe some of the physical chemists here would have some suggestions?

  39. NobodyElse says:

    Bioinformatics helped refine X-Ray guesses. How many of you try to align protein A to a related protein B and gestimate the structure? Alignment is part of Bioinformatics core function.
    Problem is that experimental part can not keep up with hypotheses generated by analysis of genomics. What a world would it be If we could automate the lab to test all of them? Than again try to automate oil change in your car.

  40. Brandon Berg says:

    Dearieme @12:
    Sort of. Once they identify the gene that causes a disease, they can create transgenic animals and human stem cells expressing that gene, for use in drug testing. While this isn’t perfect, and results from animal and in vitro testing often fail to pan out in humans, it’s a huge improvement over the alternative.

  41. Brandon Berg says:

    Dearieme @12:
    Sort of. Once they identify the gene that causes a disease, they can create transgenic animals and human stem cells expressing that gene, for use in drug testing. While this isn’t perfect, and results from animal and in vitro testing often fail to pan out in humans, it’s a huge improvement over the alternative.

  42. WAT says:

    Big data is here to stay, we have to learn to deal with it not run away.

  43. Esteban says:

    Bioinformatics encompasses a lot of data types. To the extent the author’s rant is directed toward gene expression data, I can sympathize. Utilizing split samples, I’ve estimated the assay measurement error in any single gene expression result to be roughly 2-fold up/down based on a 90% confidence interval. That’s a generalization of course, but the point is that building complex statistical models using expression data is highly prone to GIGO.

  44. DP says:

    ” is it unfair to the state of bioinformatic databases?”
    Surely that depends on the databases – from my direct experience some of the commercial ones, such as the proprietory ones developed by Ingenuity or GeneGO Metacore, are getting to be pretty decent; at least good enough that people are happy to pay the hefty licence fees. But I think heavy manual curation is the key there, and of course that doesn’t always happen with all databases, even the heavily used ones.
    I guess it comes to the old serenity to accept the shitty databases I cannot change, the courage to change the databases I can, and the wisdom to know which is which…

  45. to 12 says:

    Arrogance is not a virtue. BCR-ABL, ALK and KRAS muts are not curing they are guiding the cure.
    Taqman and qPCR are measuring gene expression, so let’s ignore all of the research with PCR and Taqman.
    Let’s drop Western as part of it as well.
    Problem is in our ability to interpret results and reporting positives rather than reality.

  46. Insilicoconsulting says:

    #Calvin,
    I was also one of those developing bacterial essential gene prediction algorithms in the late 1990’s.
    Given that one took essential genes proven by knockout from diverse bacteria in the training set, predictions made on newer genomes or genes with unknown knockout phenotype were surprisingly accurate. (as attested by a customer).
    Taking the training set as only the M. genitalium genome without the cell wall genes led to some bad models.
    I don’t think bioinformatics is any different from wet lab techniques, in terms of hype vs the delivery.
    Bad data and thus bad models is a omnipresent problem not only limited to bioinformatics.

  47. JimM says:

    “…
    The vast majority of scientists on both sides of the debate agree with the following|:
    · that climate has always changed and always will,
    · that carbon dioxide is a greenhouse gas and warms the lower atmosphere,
    · that human emissions are accumulating in the atmosphere,
    · that a global warming of around 0.5OC occurred in the 20th century, but
    · that global warming has ceased over the last 15 years. …”
    I’ve seen this last point a number of places recently, but don’t know how well-accepted it may or may not be.
    But if true, it’s hardly reassuring, because pouring heat into a system without seeing an increase in temperature is a classic signal of a critical point associated with a phase change, and we might see drastically reduced system heat capacity once we get through the critical point to whatever is on the other side (as we do in water vapor compared to liquid water, for example) — meaning that temperature would then go up much more sharply than it did during the 20th century for equal increases in heat inputs.

  48. Nekekami says:

    @38:
    Swedish Justice Department, though that does not preclude foreign pressure, given that carbon rights trade etc is a business worth billions of Euro.

  49. Anonymous says:

    1)bioinformatics have been extremely helpful for protein structure.this sounds more like some programmer pissed off that some simple algorithm didnt solve “biology” like the engineers like to tell us how we are all doing it wrong.
    2) sure as hell doesnt look like global warming stopped in the 90’s http://www.metoffice.gov.uk/hadobs/hadcrut4/data/versions/HadCRUT.4.1.1.0_release_notes.html

  50. nona says:

    1)bioinformatics have been extremely helpful for protein structure.this sounds more like some programmer pissed off that some simple algorithm didnt solve “biology” like the engineers like to tell us how we are all doing it wrong.
    2) sure as hell doesnt look like global warming stopped in the 90’s http://www.metoffice.gov.uk/hadobs/hadcrut4/data/versions/HadCRUT.4.1.1.0_release_notes.html

  51. Henry's cat says:

    I love a good ‘screw this I’m outa here’ rant. We all have moments when we want to raise a mid digit at the insufferable screwups that we work with. I salute the author for finding better employ where pastures are most likely much greener. I take issue, however, with his notion that the opposite of ‘inept’ is ‘ept’. Truly, truly shocking.

  52. Helical_Investor says:

    So a programmer is upset that biologists and such just don’t know how to communicate with him or appreciate the limits on what he can do. And that is their fault? This guy does need to ‘go where he is understood’, because he apparently lacks the ability to communicate across fields (It is hard, and not for everyone). Sounds like a guy who can ‘apply a solution’ to a problem but not necessarily create one. That is common (truer of most of us than we’d care to admit).
    I’d note that the biologists did hire him to begin with, so they at least do appreciate the need / utility of bioinformatics. So good for them for trying the push the envelope on what it can do for them.
    You know what made microarray technology ultimately practical. Some chemists (and biologists) had the foresight to recognize they needed to hire / consult with some mathmaticians to effectively extract the information they believed they could obtain via fluorescent signals of hybridization. Thank goodness they hired ‘the right ones’ and not this guy.
    Zz

  53. Helical_Investor says:

    So a programmer is upset that biologists and such just don’t know how to communicate with him or appreciate the limits on what he can do. And that is their fault? This guy does need to ‘go where he is understood’, because he apparently lacks the ability to communicate across fields (It is hard, and not for everyone). Sounds like a guy who can ‘apply a solution’ to a problem but not necessarily create one. That is common (truer of most of us than we’d care to admit).
    I’d note that the biologists did hire him to begin with, so they at least do appreciate the need / utility of bioinformatics. So good for them for trying the push the envelope on what it can do for them.
    You know what made microarray technology ultimately practical. Some chemists (and biologists) had the foresight to recognize they needed to hire / consult with some mathematicians to effectively extract the information they believed they could obtain via fluorescent signals of hybridization. Thank goodness they hired ‘the right ones’ and not this guy.
    Zz

  54. Mike says:

    The computer nerds at Y Combinator, amid all their hand-wringing about how code that only needs to be used by 15 people tends to be written inefficiently, hit on a good point (maddening though it is). They scoff at the posted salaries for geneticists to join biology labs. $40,000 for someone with a PhD? It is to scoff, apparently.
    You can’t hire some stranger on an hourly basis to write these programs. You need it to be done by someone who is on your team and understands the problem. Effectively, good bioinformatics programs can only be written by people at about eight institutions, because they can hire full-time programmers to work on a biological project. A molecular biology PI is not going to write grants that include $90,000 to pay a leet haxor. You have a team of postdocs and students and techs and whatnot, and the guy making twice as much as anyone else is the one who just sits on his ass all day? Improbable.

  55. Ricardo Rose says:

    Bioinformatics was oversold and hyped by senior management who had little idea of the foundations, in which fields it could help and how much more development of the field was left to be done so it could become a tool which could be used on a routine basis.
    This seems quiet a recurrent theme in the industry, and as our colleagues in the finance engineering sector, just sit here wondering when the next bubble will come.
    Good all scientific method doesn’t seem to be well received in certain environments.

  56. Nodz says:

    #51. Re the use of “ept” as the opposite of “intept”. I like to think its a nod to Groo the Wanderer… Can I hear a hell yeah from anyone who got that reference?

  57. Derek Lowe says:

    Good ol’ Sergio Aragones!

  58. RKN says:

    @drug_hunter,
    There’s biological information in that “hairball.”
    It would be unwise to wait until there is more & better data to develop the methods to extract and integrate that information, in order to forward answers to relevant biological questions.
    Would like to try to overcome your incredulity sometime. Drop me an email.

  59. sivakumar says:

    The title “Farewell to Bioinformatics” of the message is not correct.
    The creation of wrong tool or selection of wrong tool is because of the error of an individual not the field.
    SO, i am not supporting the title “Farewell to Bioinformatics” because it provides immense usage to the scientific world.

  60. Jonadab says:

    Putting mathematicians and biologists on the same project is basically *asking* for a culture clash.
    Mathematicians are trained to think almost exclusively in terms of what is known absolutely for certain, and they have a MUCH higher standard of proof than any other scientific discipline. When a given hypothesis has been rigorously tested tens of millions of times by hundreds of highly regarded mathematicians in dozens of countries and has held up every single time so far with no known exceptions, it is considered an important “conjecture” and regarded as an “open problem”. When a given hypothesis has been considered by thousands of highly regarded biologists in dozens of countries and they all draw the same conclusions without any actual (falsifiable) testing, it is widely regarded as a basic scientific fact.
    So yeah, culture clash.

  61. ChristianKl says:

    A lot of the bioinformatic software is open source and open to review. In contrast most of the software that used to model the climate isn’t open to outsiders.
    In contrast to bioinformatics the data isn’t as open in climate change.
    If you compare the methods of those fields I would expect the bioinformatians to do better.
    I don’t really understand why we should accept the results of those climate models when the source code and data set aren’t open.

  62. JamesMors says:

    I can understand his point and it’s partly true. Neither single “shitty data” nor an assemblage of them can be transformed into shiny results by means of bioinformatics. But should we consider this a failure of bioinfomatics?

Comments are closed.