Open Source Drug Discovery
Could you invent a drug for free?
No. It costs on average almost $1 bn to get a new drug launched. Although much of that cost comes from counting all the failures along the way, since for every 150 drug discovery projects that get started only one, 12 years later, makes it to market.
But could you invent a drug without spending any money?
Maybe. Programmers create software all the time without spending any money. Could the same open source methods and culture create a new drug? Could it harness the power of many minds and the experience of the long tail of scientists and others, and sustain that process through the long march through to the market? In their book WIkinomics, “How mass collaboration changes everything” (see their blog), Tapscott and Williams ask the same question
Having witnessed what Linux has done for software production it seems natural to wonder whether a flurry of open source activity could unleash a similar revolution in the life sciences. What if the drug discovery process, for example, was opened up so anyone can participate, modify the output, or improve it, provided they agree to share their modifications under the same terms? Could the colectiv intelligence of the life sciencs community be harnessed to enable a more coordinated and comprehensive attack on the tropical diseases that so have far stymied the industry?
We are building InkSpot Science as a platform for collaboration between scientists in any field, but my personal interest is in drug discovery, particularly in platforms and processes for doing it differently and it seems natural to ask whether we could develop open source drug discovery projects using the software tools and data storage systems that we are adding to it. On the surface, this whole idea is ridiculous and particularly the idea that for open source to develop, there can be no ip, at least in the sense of protectable and/or confidential information. To someone who has spent 25 years in drug research, the idea that the chemical structure of compounds with biological activity should be made freely available under a community license such as creative commons or science commons seems wrong. Nevertheless, that is a critical component of an open source, community led approach and I guess I can adapt.
The idea of open source drug discovery is out there and the Synaptic Leap is one site that has been set up to develop the approach for tropical disease, an obvious place to start since this is an area which is not well served because of the lack of sufficient commercial return. That means that public and charitable funds have stepped into the research finding gap as with The Tropical Disease Initiative, One World Health, Dundee University’s drug discovery and the Drugs for Neglected Disease Initiative. The idea has been bubbling around for a while with one or two enthusiasts such as Marc Marti-Renom giving some early examples as in this Google Tech talk. There’s now even a wiki for the idea. Another champion is Jean-Claude Bradley of Drexel University who links the idea to his “open Notebook” thinking, where lab notebooks are open and shared and data and ideas are open to scrutiny. It has even reached Nature in an article by Bernard Munos. In India, the new head of the Institute of Genomic and Integrative Biology (IGIB), Samir Brahmachari, has similar ambitions.
The clearest exposition that I have seen comes from publications by Steven Maurer and co-workers. His articles look at the legal, motivational and social background to open sourcing drug discovery, taking experiences from open source software development.
In all the discussion and enthusiasm for open source drug discovery, it is not obvious that there is anything happening, beyond discussion of the idea, grant applications and setting up the mechanisms to support it. Over the next few months, we will start experimental open source drug discovery projects using existing services on-line and molecular design software tools that are available to us, drawing on the growing body of open source scientific software and other systems. Rather than discuss the concept, it seems a good idea to try and do it for real, using tools we already have and adding more as we go along.
If anyone would like to help in this, get in touch.
Comments
David,
I really like the ideas that you are discussing here. Perhaps we could explore a collaboration on the development of anti-malarial compounds. We could certainly benefit from more feedback from people with computational chemistry expertise. There are currently 2 major objectives in our research:
1) Predict which compounds from our library will precipitate from the reaction conditions in pure form. See this post for more explanation.
2) Predict which compounds from our library will exhibit anti-malarial activity. Lately we’ve focused on falcipain-2 but we are open to other strategies.
This is an interesting undertaking. You maybe able to leverage it with the databases hosted by Collaborative Drug Discovery (www.collaborativedrug.com). How do you see InkSpot differing from The Synaptic Leap?
Jean-Claude: Thanks for the support and I appreciate the comments. I will read up on both your solubility problem and the work on malaria in more detail and post later on what we might be able to contribute. I am open to discussion on what targets. A target with good protein structure information and/or a lead compound where we can find existing SAR would best suit our tools. I am reading up on the semi-carbazone for Chagas literature at the moment, but Tropical Diseases is a new area for me so I have bit to catch up on.
Bernard: Inkspot Science is aimed at any scientific domain and we aim to support a wide range of software and data types. I have a personal interest in drug discovery and so that’s where I want to develop the first “inkspot”. We hope other people will lead the development of other inkspots (i.e. other science domains). I have just put in a request to register for Synaptic Leap since I very much like what they are doing and hope that some collaborations might spin off. The focus of Inkspot is to provide the tools for doing science on-line as well as those for discussion, so I would hope that what we are working on would complement the work there.
David, can only echo what Jean-Claude is saying here. Welcome to the fray and good luck with what you are trying to do. We would be very interested in talking about both collaborative tools and their development as well as the drug discovery angle.
I am working up with some colleagues some proposals to take our lab recording framework into teaching settings where students (undergraduate and high school) synthesise, test, and screen in silico compounds against a range of drug targets. This was inspired by Jean-Claude’s work and his collaboration with Brent Friesen at Dominican University on undergraduate chemistry teaching. You can see some of the initial stages of this here. All the data from this will naturally be open so available for remixing and reworking as well as possibly testing some of your tools?
Cameron: Many thanks for the welcome. I just had a long conversation with Jean-Claude and he mentioned you. Yes - by the looks of it we will have a lot to talk about. I like the idea of the undergraduate labs that give students experience of the predictive approaches and then take them into the lab. There are probably ways our systems could be used. I am planning to start blogging about them, particularly the Discovery Bus auto-QSAR with worked examples over the next few weeks. I understand there is an e-science meeting in Southampton in September?
Glad to have your input at TSL. We have a few projects needing scientists from the community. For example an organic chemistry project that’s currently active needs experts in asymmetric hydrogenation to solve a simple problem in tropical diseases.
The background to this whole general idea of open source in drug discovery is something we also discussed in an Australian Journal of Chemistry article. I think there’s a lot of science that can be done, quickly, in this mode. Whether that means one can actually develop a drug is an interesting question. That’s what we’re trying to find out with the new grant we secured. In this case we’re receiving investment from the Australian government and WHO, but it’s still a non-profit project, where an open source element was crucial to the proposal.
Maybe the question is not “Could you invent a drug without spending any money?” but “Could you invent a drug without making any money?”
Mat
Thanks for the comments Mat and I will follow up on the article.
Somebody needs to make some money at some point if the supply and delivery of quality material to patients is to be sustainable. I could speculate that with community, charitable and/or public support a drug for a tropical disease could get through discovery and clinical trials. Then the market is open for a generics company to pick it up. When we pay (say) $2 for a tablet, we are paying perhaps 25c for the physical material and $1.75 for all the information that was created in its discovery and trialling, which is protected by patent for around 7 years. If the information is free, and not protected by patents, the generics can manufacture and sell the same tablet for 40c. They make their profit, constrained by competition, the drug is much cheaper and the supply is guaranteed by the commercial incentive. Its even possible that the generics would reach in to the later stage clinical trials and take over the funding.
Congratulations on the grant!
David, we are planning a workshop on Open Science at Southampton on 31 August and 1 September. More details are here although they are rather sketchy at the moment. As Jean-Claude is over then perhaps that would be a good opportunity to meet up. Others at Southampton like Jeremy Frey would also be interested in what you are doing I think.
Actually the grant programme is looking good in this area. Mat has got one, JC has one in with Gates Foundation and I and others are trying to finalise something on the undergraduate front for Wellcome. Actually it would be great to have you in on that if you’re interested. QSAR is an obvious thing that is missing from our current mix. Drop me an email and I can send you the proposal as it currently stands.
Cameron - yes the Southampton workshop would be a great opportunity for us to all meet up. By the way, in terms of QSAR for Open Drug Development, people like Rajarshi Guha have contributed a lot, especially for our project. However, InkSpot may provide a way to systematically and automatically archive QSAR and other computational experiments in an open database. That is something that we don’t have and could certainly use.
Sorry, yes, I should have been more explicit. What I meant was that in the grant I have been writing recently (with people at RAL/Soton/Bangor) we don’t have any QSAR or much in the way of in silico analysis at all at the moment. Rajarshi’s contribution to everything that has gone on is very important. And yes I like the sound of what David is talking about here in terms of archiving and making available.
QSAR is a speciality of ours particularly auto-QSAR which we have been working on for the last few years — see this presentation at last years UK QSAR meeting. Also virtual screening by shape and pharmacophore matching.I’d be interested in participating in an grant proposals. I have a new academic position with seed funding, but now want to extend that. So grants are good!
If anyone has some data sets (structures and properties) you’d like to have us look at for a QSAR model, then let us have them and I’ll post the results. I just need a text file with an id column (e.g. UNIV001, UNIV002, …), smiles column and property column). QSAR works best with more data (60+ compounds) and a wider spread of results (>2 Log unit variation).
In context of the above discussions, it is imperative to mention that India has taken a novel approach to drug discovery through the Open Source Drug Discovery initiative (Cell, 2008) under the guidance of Prof. Samir K. Brahmachari, former Director of Institute of Genomics and Integrative Biology, Delhi, India, and currently, Director General of CSIR—a chain of 37 research institutes in India. OSDD is a CSIR-led initiative. It’s key focus is to provide affordable healthcare to the developing world by providing a global platform where the best minds can collaborate and collectively endeavor to solve the complex problems associated with discovering novel therapies for diseases like Tuberculosis, Malaria, Leshmaniasis, etc. “Open source approach” has succeeded in IT & “Human Genome Sequencing Project” (HUGO initiative) and there is an urgent call for such a programme in Healthcare. OSDD has taken a cue from this. The market size has been a major driving force in targeting human diseases for new drug discovery programs. Each such programme costs approximately between US$ 250-800 million and takes about 12 years to hit the market. Absence of a considerable market (in terms of monetary gains and recouping R&D expenses) is one of the reasons why little interest has been shown in developing drugs for infectious diseases like tuberculosis. This is substantiated by the fact that there are 399 drugs under clinical trials for cancer and only 6 under clinical trials for Tuberculosis. The OSDD concept aims to synergize the power of genomics, computational technologies, and bright young minds from Universities and Industrial partnerships coupled with a strong inclination to apply a concerted effort to bring down the cost of drug discovery. OSDD aims to discover new chemical entities and to make them generic as soon as they are discovered, thus expediting the process of drug discovery. (http://spicyipindia.blogspot.com/2008/03/spicy-ip-interview-with-dr-samir-k.html).
The present discussion raises concern about sharing data, for example, “chemical structure of compounds with biological activity” under the community license such as Creative Commons (http://creativecommons.org/) or Science Commons (http://sciencecommons.org/).
Prof. Brahmachari argues that drug discovery process should be in the public domain so that the entire knowledge about the new class of molecule becomes generic. It does not mean that one cannot patent it or protect it. One can apply for the Intellectual Property Rights too, but one can share among the open source community too. It would work as an innocentive model, instead of patents or royalty, credits and prizes would be awarded. (http://www.scidev.net/en/features/q-a-advocating-open-source-drugs.html).
With reference to the answer provided by Dr. David Leahy (“Could you invent a drug without making any money?”), Prof. Brahmachari too has similar views. However, he illustrates it by using a different example. He talks about a factory making an expensive product and also about a paddy field. A wall can be built around the factory to restrict access, but a wall cannot be built around the paddy field as it would raise the cost substantially.
Open source applies to the scenario illustrated by the paddy field analogy, not the factory analogy.
Prof. Brahmachari sees this as a war – a war for the right to health and affordable medicines for all. Affordable healthcare is a global problem. It is as big as the problem of terrorism.
David…what you are doing sounds very intriguing and there might be an opportunity to tap into the resources that we are building over at ChemSpider (www.chemspider.com) to help in your efforts. We are starting to aggregate experimental data associated with chemicals and these data could eventually feed into your QSAR models for example. We have recently demonstrated that we could start to host “Notebook” pages and show here one of JC Bradley’s UsefulChem pages: http://www.chemspider.com/Chemical-Structure.10481766.html. We have dabbled in “virtual screening” using the LASSO approach (http://www.chemspider.com/blog/announcing-the-chemspider-ligand-activity-project-partnering-with-simbiosys.html). I think there might be value in us connecting.
Man thanks Jyoti and Anshu for telling us more about the initiatives at CSIR. I had only seen the article and a few quotes from Prof. Brahmachari, so it is fascinating to find out more about what is behind the headlines. Please keep us updated with your progress and if you see opportunities for collaboration at some point in the future please get in touch.
Anthony — thanks for getting in touch and would be keen to start a conversation with you. I am very familiar with what you have been doing with ChemSpider and have used the site. I’d be particularly interested to know more about your plans for structured data collation that we could apply our automated QSAR approaches to.
Our team after completing an extensive survey on Bioinformatics products market has found that it is extremely important to engage biologists in the computing side of bioinformatics and for biologists to describe biological problems in a way accessible to computer scientists if Bioinformatics has to be a successful industry.
Bioinformatics is the future… lets develop and validate the drug-like targets, small molecules that will allow scientists to further do their R&D in biotech and pharmaceutical companies in the area of unmet medical needs under this OSDD program.
The world field is going to change dramatically in the next five to 10 years and its the small companies that are going to drive the change.
Write a Comment