Skip to Content

Magic Open Source Savings Await

I’ve never known quite what to make of the “open source pharma” idea. I know that there are a lot of people working on things in this area, and that there’s a lot of interest. But as someone who does drug discovery for a living, I can’t quite see how it can work. And I’m not sure if that’s just because I know a lot about the subject (and I’m therefore more realistic about it) or if it’s just because I know a lot about the subject (and I’m therefore stuck in my standard ways of thinking about it). Too much of the talk in this area seems to be confusing it with open-source software, without realizing that any such confusion is possible, and seems to devolve into vague generalities far too easily.

A reader sent along this link, a blog from a legal firm specializing in standards-setting and technology consortia. The author just got back from a conference on the whole open-source-pharma idea, and seems to have absorbed the general tone of the meeting. And that tone is. . .well, I have to say that it’s not completely wrong, but it’s not quite in synch with reality, either. There’s a lot of talk about how most of the innovative work is done in universities, and how there are “hundreds of promising discoveries” that never get properly looked at by Big Pharma, and how there are so many drug repurposing possibilities that no one is even bothering to look at, and so on.

This is too simple a view of the world. For one thing, it seems to posit that there are a bunch of dreamy-but-innovative academic labs, and then there’s Biiiig Phaaarma, who buys up their innovative innovations. Not much in between, apparently – there’s a passing reference to small companies being bought up, but that’s just in the context of all their good ideas being smothered. But a lot of these small companies come right out of university research, and there’s this whole mechanism called “venture capital” to get them going. Some of that VC money, in fact, is coming from the big drug companies themselves these days. It also ignores the long history of companies, academic labs, and even the NIH in trying to repurpose known drugs and abandoned projects. When someone comes along talking as if this is an idea that’s just never been tried before, it makes me suspicious about what else they don’t know.

Then there’s this:

Proprietary practices have other pernicious effects as well. Multiple pharmas may be exploring the same drug possibilities at the same time. Worse, one company may have already learned that the line of inquiry is a dead end, either because the animal trials did not replicate in humans, or because of toxicity. Similarly, where research is not published until the patents have been filed and the drug introduced, a decade or more can pass before other researchers can benefit from and build upon it.

Ah, but if some drug wipes in the clinic, other people know about it. We actually keep pretty close tabs on each other in this business. Companies tend to announce it when an interesting program goes into clinical trials (investors and all, y’know) and the advent of has formalized that process. And they most certainly do announce it when that clinical trial fails or succeeds (investors and all, y’know), and everyone finds that out long before the paper gets published. You’d also think that a technology-focused law firm would realize that other researchers can even benefit from patent filings as well as journal articles.

And it’s pernicious that “multiple pharmas may be exploring the same drug possibilities at the same time”, is it? Anarchy! That is a feature of drug discovery, folks, not a bug. That’s how things get discovered. You may note – if you take a few minutes to look into it – that most of the time, the majority of these programs don’t actually work, but one or two make it through. Which one or two, though? Can you tell at the beginning which companies will succeed? Not at all. That’s why having several investigating the same area is actually a good thing, for the most part. That phrase “the same drug possibilities” also makes it sound as if these multiple companies are pretty much working on the same compounds, but that’s not what happens, either. People have different chemical series, with different advantages and different liabilities, and that’s most certainly a good feature, too, given the state of our predictive abilities.

Here’s the part where the post really starts talking about what might replace our current system:

But imagine, if you will, if we were to map out the entire drug development process from initial theory through research, clinical trials and regulatory approval (and there are many, many sub-steps along the way). After we’ve produced our detailed work flow model, we can then spec out the type of IT platform needed to support that end to end process, and also the type of database needed to hold and share all the resulting data.
Of course the software tools we will want to use will insure that all of the data from one end to the other is in compatible formats to maximize efficiency in developing that drug, and also to permit maximum universal use of the resulting database by other researchers. The software tools comprising the resulting framework will be best of breed open source tools (many of which already exist), and to the extent that there are gaps, existing tools can be optimized, or new ones created.
The work flow model will also specify exactly what number of individuals, with what skills, are needed at each step along the way. It will also identify the points at which crowd sourcing of knowledge from appropriate experts would be helpful, as well as what types of funding might be appropriate and available at which junctures. Importantly, it can also be designed in such a way as to dramatically drive down the costs at every stage, from beginning to end.
It’s that last sentence that really loses me. Up until then, I’m thinking “OK, this is what a drug company does. . .yep, that’s what a company does, too. . .” We’re trying to make sure that it’s other people inside the same company who can use the data, instead of everyone else, but the problems are the same. But where are those dramatic cost savings coming from, at every stage? The only thing I can imagine is free labor, somehow, because we’ve tried everything in the paid-but-less-paid line, that’s for sure. I think that’s what the crowd sourcing part must mean. And while donated labor can work in software, there’s not a good way to scale it for drug discovery, in much the same way that while open source software is a big thing, open-source chip design and production isn’t.
Basically, this seems to be saying that there’s a way to make the whole drug discovery and development process “dramatically cheaper”, but that the people who are actually spending their own cash to do it have somehow failed to realize any of these savings, because Proprietary. And the savings will happen at every single stage of the process, too, because Software. And while I don’t rule out some open-source development, and I’m encouraging people to give it a try to see what happens, I’m not going to be holding my breath waiting for it to be just so wonderfully disruptive and better than everything before.

48 comments on “Magic Open Source Savings Await”

  1. Old Timer says:

    This is disturbing. Many truly smart individuals are going into tech. Unfortunately, their success working within the bounds of human knowledge blinds them to the realities of working outside the bounds of human knowledge. Are they not required to take science courses anymore? Or have university science courses been watered down so much people don’t understand this? I’m afraid this problem is going to get worse, thereby negatively affecting science in the US.

  2. Hap says:

    It’s a lot easier to expect people to contribute time without pay when they have less investment in that time – if you can start working from high school or before rather than after ten or so years of school, you don’t have as many fixed costs to pay off with that time.

    It seems like, between the emphasis on open source (which requires someone, at some point, to be paying people for their work, which probably isn’t all that different from the work for which they aren’t being paid), the need for lots of cheap, preferably indentured labor (their push for lots of new visas while laying off higher paid domestic labor), and the desire to water down patents, that a lot of tech and software businesses aren’t going to be sustainable. If you think that everyone else’s work should be cheap or free but yours should be compensated (generously), well, either you’re not going to be around long or you’ll be a very unpleasant master.

  3. LeeH says:


    I do think there are some good ideas in what you described. The dream of having a standardized format for drug discovery data, and having it accommodate not just early discovery (mostly pharmacology) but in-vivo and PK data is a noble one, although perhaps too gargantuan a task for the open-source model. And having companies actually share this expensive data is also a pipe dream, aside from the technical aspect of the data sharing and the technical hurdles for the visualization (you’re spoiled, having perhaps the most integrated system on the planet).

    A possible solution to this problem is not necessarily a technical one. I believe that the sharing of data is greatly impeded by how publications are crafted, that is, that they are too formal and heavy and require far too much work to prepare. If there were a more casual way of depositing experimental data, the lessons learned by everyone could be disseminated far more quickly and ultimately distributed in a more structured way by public databases such as chEMBL (which is admittedly public, but exists by virtue of a huge amount of money donated by the Wellcome Foundation). The net effect would be similar.

  4. Matthew Todd says:

    Lots of excellent points, Derek, thanks for featuring. The meeting held involved 3 days of talking about the issues you raise, and there’s a paper under review that tries to address these. The two things I’d like to mention quickly up front, though, are: i) developing medicines is complex, so adequately assessing the impact of open source (i.e. zero secrecy) on that process will also be complex so we can’t necessarily expect a single blog post, intended for pharma outsiders, to address everything, and ii) traditional/big pharma, or rather more generally the private sector, should be viewed as potential collaborators in open source projects, not adversaries (apologies if I’m misreading your subtext). One of the most powerful stories at the meeting was that of fexinidazole, a compound that is likely to clear clinical through a mutually beneficial collaboration between Sanofi and DNDi. If something is open source then anyone can work on it, and as we know from software (and in my experience in running open source drug projects) the private sector are enthusiastic collaborators and supporters of the approach, often for reasons of self-interest.

  5. jbosch says:

    I thought BIG Pharma had agreed to share their new data with ChEMBL to allow academic users benefit from their wealth of information before heading down a dead end.
    I vaguely recall an announcement from GSK, Novartis (?) and Merck (?) about 2-3 years ago about this.

    Regarding OpenSource Drug Discovery, there are many efforts ongoing in academic labs where people donate their time to help a project with their particular skills. This ranges from computational to medicinal chemistry and actual synthesis.

  6. ex-sanofi says:

    Unlike open-source software, it is difficult for a sixteen year old kid, working in his bedroom, to make much of a contribution to pharmaceutical research.

  7. Vader says:

    Open source seems to work pretty good for software. It’s why I have Ubuntu on my home system.

    In other contexts? This seems to apply:

  8. Grey Williams says:

    So you’re not a big fan of the “because Software” argument? It’s so obviously true, though, don’t you see, because History!?!

  9. David Cockburn says:

    I do find it surprising when people swallow the open source pharma story whole. I can certainly see learning some lessons from software development and adopting universal data standards all the way from discovery through clinical trials, though that may not be as easy as it sounds given the time scales. But do people really believe that there are lots of brilliant ideas coming out of academic labs which can be just bunged into the clinic to reappear soon as marketable drugs? Or is this just anti-pharma propaganda?

  10. milkshake says:

    discovering and developing a drug is far more like producing “Best Picture” Oscar-winning movie, than writing software. The self-confidence of open source drug discovery promoters will not survive encounter with reality. What a self-indulgent waste of time and effort.

  11. Jack Scannell says:

    I was invited to the 3 day meeting from which the blog post came. It was my introduction to the concept of open source pharma. At the meeting, I saw various grizzled medicinal chemists. However, I saw no spotty 16 year old programmers.

    I think it is clear to most people who look at the drug industry that the intellectual property tail often wags the therapeutic dog. The folks at the meeting were, in general, trying to put the dog back in charge of its tail. There was some missionary zeal on show; perhaps a spill-over from tech. But the meeting was tempered with plenty of pragmatism (e.g., the Structural Genomics Consortium expanding the “pre-competitive” space to rationalize the way kinases are characterized). Furthermore, much of the meeting focused on diseases of the poor, with contributions from organisations such as DNDi, MSF, CSIR-OSDD, and MMV. For these organisations, the therapeutic opportunity is large but the commercial opportunity is negligible. Here, there did seem to be an obvious opportunity to push data sharing and “crowdsourcing”; sometimes to re-purpose old drugs at very low cost; all without torpedoing intellectual property and commercial investment.

  12. Kelvin says:

    It’s very simple. We are all limited by the same pool of knowledge and understanding of human biology and disease on a general level, and in particular by the overall quality of that information. As time progresses, we pick the best opportunities (with the best expected returns) first, leaving poorer-quality opportunities with diminishing returns for later. That will always be true, regardless of where the opportunities come from, except that if you do manage to get better ones from elsewhere, you have to pay fair value for them. Basically, you only get what you pay for (and vice versa, except by luck) based on the quality of information, and that will always diminish over time as we have seen recently with the quality of published research, as well as overall return on R&D over the past 60 years. Bottom line: if you want to create value, you have to create it yourself by doing something that nobody else has done before; you can’t create value by buying something that already exists, or by duplicating what has already been done.

  13. Eric S. Raymond says:

    I was the foundational theorist of open-source software development back in the 1990s, and have received a request to respond to your post on open-source pharma.

    Is there misplaced idealism and a certain amount of wishful thinking in the open-source pharma movement? Probably. Something I often find myself pointing out to my more eager followers is that atoms are not bits; atoms are heavy, which means there are significant limiting factors of production other than human attention, and a corresponding problem of capital costs that is difficult to make go away. And I do find people who get all enthusiastic and ignore economics rather embarrassing.

    On the other hand, even when that idealism is irrational it is often a useful corrective against equally irrational territoriality. I have observed that corporations have a strong, systemic hunker-down tendency to overprotect their IP, overestimating the amount of secrecy rent they can collect and underestimating the cost savings and additional options generated by going open.

    I doubt pharma companies are any exception to this; when you say “the people who are actually spending their own cash to do it have somehow failed to realize any of these savings, because Proprietary” as if it’s credulous nonsense, my answer is “Yes. Yes, in fact, this actually happens everywhere”.

    Thus, when I have influence I try to moderate the zeal but not suppress it, hoping that the naive idealists and the reflexive hunker-downers will more or less neutralize each other. It would be better if everybody just did sound praxeology, but human beings are not in general very good at that. Semi-tribalized meme wars fueled by emotional idealism seem to be how we roll as a species. People who want to change the world have to learn to work with human beings as they are, not as we’d like them to be.

    If you’re not inclined to sign up with either side, I suggest pragmatically keeping your eye on the things the open-source culture does well and asking if those technologies and habits of thought can be useful in drug discovery. Myself, I think the long-term impact of open data interchange formats and public, cooperatively-maintained registries of pre-competitive data could be huge and is certainly worth serious investment and exploration even in the most selfish ROI terms of every party involved.

    The idealists may sound a little woolly at times, but at least they understand this possibility and have the cultural capital to realize it – that part really is software.

    Then…we see what we can learn. Once that part of the process has been de-territorialized, options to do analogous things at other places in the pipeline may become more obvious,

    P.S: I’ve been a huge fan of your “Things I Won’t Work With” posts. More, please?

    1. Zach says:


      I read your writings on open source software back in the day. They were very interesting, but I wonder if they don’t make economic assumptions which are true for software and less true for natural sciences.

      Economically speaking, software
      1) Can be created by a small motivated team
      2) Has very small costs of reproduction
      3) Can be reused in a meaningful way by a large group of people with similar needs
      4) Can be tested straightforwardly and cheaply
      5) Requires only a certain amount of upfront education to make useful contributions
      6) Faces minimal regulation in most contexts

      In contrast, applied science
      1) Usually requires larger teams and expensive equipment
      2) Has large costs of reproduction — even “generic” drugs require billion dollar plants
      3) Different teams might require specialized knowledge, such as specific chemical synthesis techniques
      4) Is very expensive to test
      5) Requires lots of education and training to make a useful contribution
      6) Faces a strict regulatory environment.

      As I see it, all of those differences point toward a need for strong intellectual property and a reliable revenue stream for natural science research that isn’t necessary for software development.

  14. John Dallman says:

    As someone who works in software, and reads this blog for fun …

    The trick of all this is the ability to test what you produce. Open Source software development works really well when test cases are comparatively easy to produce; the GNU Compiler Collection is probably the world’s single most successful open-source project, and an enabler for an awful lot of others. And creating test cases for a compiler is a matter of writing some more code. Knowing what to write isn’t simple, but if you’ve had your head inside the compiler for a while, you’ll know how to test at least some aspects of what you’ve changed.

    Testing your ideas for drugs … that’s more expensive. Some of it is computational, but most of it needs living tissues, or whole animals, as best I understand it. That’s quite heavily regulated, and also presumably needs careful standardization of the testing, most easily accomplished by doing a lot of it in one lab. Having lots of small labs each testing parts of a range of doses … doesn’t sound like a good way to get reproducible results.

    A comprehensive and trustworthy method of testing drug candidates with software would change things utterly, and might well make open-source drug development meaningful. I get the impression we’re some way from that, although if someone wanted a big and meaningful software development project, it’s there waiting for them.

  15. Peter Kenny says:

    I do see advantages of what many refer to as an ‘open source’ model for drug discovery activities that are funded by taxpayers or charities (20 years from now this may be the case for most drug discovery activities). However, as soon as investors become involved the issue of ownership rears its ugly head and ownership is especially complex given long development times for drugs, IP ambiguities and uncertainties as to whether the project will even lead to a product. Personally, I don’t think the software term ‘open source’ is appropriate in this context and believe that ‘open drug discovery’ would have given the initiative more credibility. Some of the propaganda along the lines of ‘we can do drug discovery like we developed linux’ comes across as a bit naiive and tends to devalue the initiative

  16. Shanedorf says:

    @ John Dallman:
    “A comprehensive and trustworthy method of testing drug candidates with software would change things utterly, and might well make open-source drug development meaningful.”

    Read up on the Simcyp Consortium to get a little idea on what’s out there right now. It only covers one arena of drug development, but its a start in that direction

  17. emjeff says:

    More pie-in-the-sky ideas from people who have never actually developed a drug. Their first premise, that academia has scads of great drugs just laying around, is laughable. They don’t, sorry. The other ideas are just as funny – the idea seems to be that if we just planned better, things would go just swimmingly. Uh, no, because things happen while walking down the drug development pathway, and your plan might end up in the crapper.

  18. CMCguy says:

    While I too am skeptical that open source pharma would ever be truly workable or even have strong influence on lowering costs, it would be great if there was established wider and more formal dissemination of negative data on chemistry or compounds that do not work. I know this topic has been mentioned occasionally in this blog (including suggestions for creating a journal of dead-ends or such type) to overcome where most scientists, particularly in industry in general, don’t tend to often report details regarding efforts that did not go where intended (except maybe as filler in a thesis?). The majority of papers focus overly on presenting the positive outcomes of the research with only passing or no mention of things attempted that failed to one degree or another (which my observations lead me to believe can be significant portions of all projects). Such elaborate knowledge would indeed benefit for all doing follow-on or related studies whereas now unless happen to connect with someone with the detailed info on what was done in total to warn you off of blind alleys I wonder how much duplicative rediscovery of poor results occurs (and could be an argument in favor of cost reduction). On the other hand on occasions a few pieces of negative info could inspire someone to approach from a new angle to achieve or work around a solution. This is that fun vs. frustration battle that is part of progress in R&D.

  19. Thomas says:

    The idea of “open source” is to share the material which is not a competitive advantage. Does Sony win because they have a better Linux kernel? Or a better compiler? No, but if they can shave the costs that is fine. and the culture of mutual sharing can accelerate development.

    There is a Chinese variant named Gongkai – look at bunniestudios’ website – where the sharing is in a closed circuit. Only stay in if you share back.

  20. respisci says:

    I’ve migrated from the drug discovery side into clinical development and from industry back to academia/hospital clinics running clinical trials. Most academics, even the medical doctors who are PI on the clinical trials, know little of the full drug development process. Seriously, they ask what is CMC? So when there is discussions about identifying the work flow and gaps, it is imperative to have knowledgeable and experienced people around that table. In my time with clinical studies, I have seen some poor and some truly bad study designs. However, there is no perfect cookie cutter clinical study design. You need to have the expertise with the drug in question and with the specific disease indication.

  21. qetzal says:

    I have no reason to doubt that an open source approach, including standarization of data formats, could offer significant savings. But if the organizers of this conference are really aiming to “dramatically drive down the costs at every stage,” I think they missed a key step: understanding what the biggest cost drivers are right now.

    It’s not IP, or lack of IT, or closed-source systems. It’s really just two things.

    The first is the vast amount of in vivo data, both animal and clinical, that is needed to get a new drug approved. The second is how incredibly bad we are at identifying which drugs are worth developing. And be “we” I mean everyone involved in drugs, including pharma, biotech, academia, regulators, etc.

    For a single successful drug, a company must typically spend a few $100M conducting the clinical studies to get it approved. And for every 10 drugs that enter clinical testing, only about one will be successful. The other 9 will fail utterly, but only after having spent tens to even hundreds of millions on each.

    Being able to test drugs in silico would definitely fit the bill, but I think that’s a multi-decade effort at minimum. Other than that, I fail to see how IT or open source approaches can provide anything other than incremental benefits. They don’t address the fundamental cost drivers.

    Which isn’t to say that the incremental benefits aren’t worth going after. They may well be. And I think Eric Raymond’s point about the irrationalists counteracting the naysayers has a bit of merit. But it would be nice if the irrationalists were at least being wildly optimistic about fixing the right problems.

  22. B. O. R. Ingold-Phart says:

    Those that can, do, those that can’t, devise work flows…

  23. Kyle Jansen says:

    I’m a software developer who frequently uses (and infrequently contributes to) open-source software, so maybe my perspective will be useful. Having read your explanation, I agree that open-source pharma won’t work (with current methods), but I can explain why people think it should work, and what needs to change in order for it to work.

    Open-source software works because almost all of the costs are in “R&D”, or rather just the “D”. Copying software is so cheap that it may as well be free. Since all of the cost is in R&D, the most natural economic model is to pay just for development – companies that use open-source software will hire a developer (of which there are many) to code up whatever new feature they want, or fix a bug that particularly hurts them. They then, typically, release it, because it costs nothing to do so, and garners some nice community effects. A better product will attract more developers to be familiar with it, so the cost to add even more features goes down; a better product is more widely-used so you can hire more employees trained in using it, etc.

    People think open-source pharma will work because it shares that “most of the cost is in R&D” feature. Sadly, it doesn’t even fit that completely – I can distribute a million copies of the Linux kernel for the cost of a large soda, but just the shipping costs on a single prescription bottle of drugs is more expensive than that. Synthesis is also a non-trivial cost. But the real reason open-source pharma fails is because it lacks the rest of the features that make open-source software work.

    To make open-source pharma work, you would need three things (at least as far as I can tell, I may be missing some):
    1: Make production of known compounds essentially free. Perhaps Burke’s synthesis machine, or something along those lines, could do this, if you can make it work cheaply at scale, for absolutely any compound that can physically exist, but I have doubts it’ll ever happen. Basically, medication needs to be free not because of any government mandate, but because the cost of actually charging money for it is greater than it’s actually worth.
    2: Lower the barrier to entry to the point that someone can conceivably develop a drug all the way to approval, using only common household items and training obtainable at a good high school or a regular community college.
    3A: Cure all common diseases well enough that the only ones left are flukes affecting only a handful of people, requiring customized treatment. Failing that:
    3B: A way for people suffering from a given disease to collectively and directly fund development of a treatment or cure.

    Out of all of those, the only one I think is even remotely possible in the near future is 3B. 2 might be possible if we can develop perfect computer models of the entire human body down to the molecular level, and have powerful enough home computers to run those simulations. I’m not sure Moore’s Law will hold that long, and if it does, I’m not sure this argument will even still matter. If we had such simulations, we’d trivially be able to automate a brute-force search of every conceivable compound, and so human involvement in drug discovery wouldn’t even be necessary anymore. 1 will probably be a problem for the foreseeable future, simply because the raw materials still have non-trivial cost. Even if you could “print” the perfect compound using nanomachines, stocking your printer with the seventy-odd stable, reactive elements will still cost something.

    So yeah. Open-source pharma will probably never be a thing, because if we have the technology to make it work, we won’t need it anymore.

    1. EdM says:

      I believe Kyle missed an important 4th feature:

      Testing must be nearly free.

      Most open source software is not tested for safety, and is not heavily tested before release in general. Software that has drug-like safety requirements (e.g. flight control software for commercial airliners) is rarely open source, and is much more heavily tested. The testing effort and cost for safety-critical software is several times the R&D effort, and results in product cycle times for aircraft electronics that are closer to drug development times than open source software release cycles.

  24. JC says:

    I followed the above post for the first three paragraphs and then you lost me.

    Especially this part

    ‘Basically, medication needs to be free not because of any government mandate, but because the cost of actually charging money for it is greater than it’s actually worth.’

    Back to software, why is a copy of MS Office $99 then?

    Maybe I have a different mindset than the Andy Groves and other software folks having slogged away at the bench. Maybe it’s the fumes.

  25. Severin says:

    I’m a computational chemist and a big proponent of open source data. Chembl and drugbank are invaluable and sustainable. Scifinder/Belstein and all publications should and could be publicly supported and open source.

    In terms of pharma collaboration my motto is don’t share the identity of leads and chemical libraries, but instead qsar models and descriptors. It has the double advantage of concealing a compound collection and finding out if qsar just interpolates. If it does people will converge on chemical series earlier than we do anyways with patents and at worse call out the algorithm. If we make a mistake and share too much let there be consequences, but at least we should try. At the GRC it was mentioned that we aren’t really dealing with big data yet in computational chemistry. We should be.

  26. DrSnowboard says:

    I am liking the informed discussion going on here, the balance between ‘ you can’t do that because…’ and ‘ what if we could do this…?’
    As an ex-grizzled medicinal chemist now project manager, I’m going to counter my previous comment (about the LO dashboard graphic) and stick my hand in the air for greater data sharing , in as simple a fashion as possible. Think of the advantages you get when your files aren’t held on individual laptops but in a common cloud system? No more mutant versions, no more editing an already out of date file…etc. This is happening within large companies because they recognise there are benefits, the trick is enabling the environment for between companies and between other parties – hence as mentioned the greatest success being in neglected diseases. After all, big pharma streamlined their clinical submission processes to allow fast registration and easy dossier compiling because they knew it would save them money. There was an initiative years ago at GW/GSK trying to impress on people that the output of a DD programme was a pile of information, not just a compound in a bottle, and capturing that information along the way to be able to re-use it was so much more efficient.

    1. Thomas says:

      Basically what is shared is the data that is not of specific value to the company itself, but to progress as a whole.
      With software, copyright and patent pledges are used to oil things. Some open source is freely licensed (copy and use at will), but others are licensed such that derivative works must also be open source and source code must be shared. Nowadays there are also patent pledges/pools. “We license these patents to anyone for free on the condition that you do not sue us in these fields.” Multiple companies are adding to those pools – and I could see it work in pharma for example by licensing for a specific use. But then this is also just a way to work around the software patent mess. Look up “openinventionnetwork” (one word).

  27. Design Monkey says:

    Kyle Jansen says:
    …1: Make production of known compounds essentially free. Perhaps Burke’s synthesis machine, or something along those lines, could do this, if you can make it work cheaply at scale, for absolutely any compound that can physically exist, but I have doubts it’ll ever happen.


    Burke’s machine itself theoretically could be made comparably as cheap as washing machine or printer, but the feedstock for it – very special chemicals as building blocks – are not going be cheap or free. What’s the current gold price? Something around 1100 USD per ounce, 36 USD per gram? Med chemists routinely have to buy and use chemicals, that cost more than that, but contain no precious metals. And fancy precious metal catalysts – they are even way more costly.

    Maybe after a lot of decades of refining (as current computers are refined versions of Atanasoff-Berry clunker) they will be capable of greater diversity of chemistry and will work on coal and grass clippings as starting materials, but that’s not going to happen especially soon.

  28. Wage_Slave says:

    To all of those software types out there who predict in silico screening, I have a much simpler challenge to get you started.
    Please write some code to predict optical rotation from a given structure, we know lots about light and lots about molecules; but it seems that no one can even tell me the sign of rotation let alone a magnitude!
    There are no fuzzy biological bits here – therfore I expect your answers, shall we say, by the end of the year?

    1. John Dallman says:

      To all of those software types out there who predict in silico screening,..

      I’m not predicting software screening. Just pointing out that something like it seems necessary to get a development cycle like that of open source running.

  29. JT says:

    I think there might be considerable savings in marketing/sales departments with a different model. Although I doubt an open source model will lead to this: probably it would make those budgets bigger due to competition.

  30. Bernard Munos says:

    As many posters have pointed out, the need for clinical trials precludes the open-source software model to serve as a template for open-source pharma. However, there are many ideas from open-source software that apply to pharma and can be used to design effective open-source pharma models. In one of the earliest papers on open-source pharma published nearly 10 years ago (, I argued that drug discovery activities lend themselves to open-sourcing, while drug development is better handled through outsourcing. However, a combination of the two can produce alternative drug R&D models that can develop drugs more economically. One should also note that the open-sourcing/outsourcing idea has found its way in various novel drug R&D models such as virtual companies, public-private partnerships for neglected diseases (e.g., Medicines for Malaria Venture or Global TB Alliance), and the multiple open-source portals that adorn big pharma websites.

  31. Hap says:

    I was the foundational theorist of open-source software development back in the 1990s, and have received a request to respond to your post on open-source pharma.

    How does one do that exactly? “Hello. I need some close air support. Can you send the Pacific Fleet’s carrier air wings over here right away? Thanks.”

  32. CMCguy says:

    B Munos you indicate “drug development is better handled through outsourcing” and imply such is a more economical R&D Model. Since the article cite was published almost 10 year (and is behind a pay wall so can not read details behind that) I wonder as your post suggest you still retain that argument after seeing “progress” by implementation of such a model? I have long felt, and even observed, that outsourcing R&D often creates a false sense of reduced expenditures because even though can shift efforts to locations with cheaper labor or other overhead there are invariably sacrifices in quality, time schedules and additional management requirements that can equilibrate budgets or overrun the simplistic costs model. Because activities can show up in different accounting buckets the achievements of less costs are still proclaimed. It really is a bad case of comparison to an older largely broken systems of consolidated fully capable organizations to a fragmented approach that can work but is by no means smooth or efficient to discover and develop new medicines.

  33. Matt says:

    Please write some code to predict optical rotation from a given structure, we know lots about light and lots about molecules; but it seems that no one can even tell me the sign of rotation let alone a magnitude!

    I’m not sure if this is written from the perspective of someone who has never calculated optical rotation properties from quantum chemistry software, or from the perspective of someone who has done the calculations too many times and is frustrated that you can still get the wrong sign (methyloxirane?). “Please write some code to predict XXX… where the prediction is never wrong” is an awfully high standard.

  34. MolecularGeek says:

    There’s been a lot said already about what an “open source” drug development model would look like, and people more informed than I have already made most of the relevant points. I would like to draw a bright yellow outliner box around Thomas’ comment about collaborating on things that are needed, but do not provide a competitive advantage. I’ve attended seminars on precompetitive collaboration where companies share biological information to reduce the risk that any of them end up working on yet another dog of a target that turns out to be intractable. There are also alliances (Pistiolla, among others) that share tools and models for necessary infrastructure. The end result of these initiatives may not be as world shaking as certain software moguls would hope, but they are happening and they are making positive changes in the way the industry works.

    1. hypnos says:

      I second that. In the current environment, I don’t think that true open source drug discovery will happen any time soon – except for some very exotic cases.

      However, many big companies see that they need to collaborate with each other, with smaller biotech companies / CROs having special capabilities and also academia to solve their problems and to improve their success rates. In such collaborations, questions of IP and confidentiality can be solved by signing contracts. Open data standards and common databases / platforms would certainly be of a great help to foster collaborations of this type.

  35. jbosch says:

    One thing that has not been touched upon is the resources of brains and ideas.

    I have to admit, I have no industry exposure as in I never worked for pharma, but do have good ties to couple of people.

    A benefit of Open Source Drug Development is that many people can contribute to a specific tailored project and contribute their ideas and brain power. I believe a lot is gained in the conception phase of a project where money is essentially irrelevant. As pointed out earlier, neglected tropical diseases are a good starting point as a testbed to see if this may be a feasible strategy.

    Just think about that space craft that was thought to be impossible to develop within the price range. Sure there were some serious setbacks, but if we don’t try it, we might as well stop doing anything.

    Innovation does not happen in a closed environment.

  36. Bernard Munos says:

    to MolecularGeek and hypnos
    Your criticism is well put. However, my view of outsourcing is not so much about shipping work to China than tapping the idle capacity that exists in the system, including in the US. At any time, there are labs, CROs and other facilities that have idle capacity because a contract ended, a study terminated early, etc. When that happens, some of them at least, are willing to offer their services at marginal cost instead of the fully-allocated costs typically paid by pharma, because it helps pay bills. That’s a BIG difference that some companies have astutely leveraged to their benefit. I’ve heard remarkable stories from some of them. You also have companies such as Assay Depot that have been created to help drug companies tap this excess capacity, and they have compelling stories of their own. (Disclosure: I am on Assay Depot’s board of directors.) This may not help run a 16,000-patient trial, but it will help run all kind of lab work without having to invest in bricks-and mortar, wait in line for in-house capacity, generate bogus business cases, etc.

    1. hypnos says:

      @Bernard Munos: I can imagine that some organisations would be willing to donate e.g. their unused slots to some charity working on – say – an under-researched orphan disease.

      However, I’m very sceptical about your “idle capacity” argument since I believe that this is exactly where real innovation takes place. Idle capacity means that people have the time to take a step back and think and discuss about what they are doing and how it could be done better. Many great ideas were born on a friday evening when two guys were waiting for their experiments to finish.

  37. Bernard Munos says:

    Just ran into an appropriate quote from Novartis’ Quotable Science page on Pinterest (, It quotes Jeff Howe (from Wired) who sees open-source at “the process by which the power of many can be leveraged to accomplish feats that were once the province of a specialized few”. Certainly applicable to drug discovery.

  38. Jack Scannell says:

    For an “open science” flavored description of the discovery of the world’s most lucrative drug class (anti-TNFs in inflammatory disease) see their discoverer, Marc Feldmann, talking at the 2015 Goldlab Symposium: . The talk was titled “Can We Define a More Cost-Effective Path to Better Therapy? Reminiscences from the Front Line.”

    As far as I can tell from the video (although the resolution is not terribly high), Feldmann is not a 16 year old computer geek.

  39. Garrett Wollman says:

    Just wanted to comment on one of Derek’s points that hasn’t been addressed by anyone else: the utility of patents for information sharing. I would not expect too many people from the software world to even consider this a possibility, because in the computing field, patents are written in deliberately impenetrable, obfuscatory legalese that experienced practitioners are often at a loss to make any sense of, let alone learn useful techniques from. I wouldn’t expect the same to be true for a composition-of-matter patent issued on a newly invented drug — ultimately, you have to claim the actual structure of your compound, no? (Or the actual process involved in its manufacture, etc.) In the software world, the conference presentations, papers, and most of all, the actual source-code-as-published-in-github are the knowledge transfer mechanisms of choice. (That’s not even getting into the issue that many software companies prohibit their employees from even LOOKING at patents, lest they get done for willful infringement and treble damages.)

  40. biff says:

    “multiple pharmas may be exploring the same drug possibilities at the same time”

    I understand that multiple (academics) may be exploring the same (research) possibilities at the same time.


  41. Excellent beat ! I would like to apprentice while you amend your website, how could i subscribe for a blog site? The account helped me a acceptable deal. I had been a little bit acquainted of this your broadcast offered bright clear concept

Comments are closed.