We are now in a phase where it takes millions of USD to bring a drug to market. Other than the fact that low hanging fruit are gone and more science needs to be done for discovering efficacious molecules, other problems are related to regulatory hurdles and costs of clinical trials. In a general sense, the regulatory oversight has indeed helped bring toxicity of drugs into focus. But it has also meant that garage science and garage based startups cannot realistically hope to find a cure for diseases. This is quite unlike the tech scene where the investments required in terms of infrastructure are low and ideas and teams are more important.
So its when some university or other researcher elucidates a terribly critical disease pathway and shows that some molecule can address this does he/she get sufficient funding. Even so, many such startups never are really able to bring any drugs to market. Personally I have known several companies that have potential drugs for cancer but cannot dream of bringing them to market.
Can we, should we think of an open source model for drug discovery then? Other luninaries have certainly thought of this before me, but let me contribute my tuppence worth. Why keep something in your head if it can be blogged for posterity to bang their head about: -). Note that I have been thinking of a for-profit but open source model but it can obviously apply to a not-for-profit organizations too.
For what its worth here are my ideas about open source drug discovery , in a series of questions and answers.
Open Source Model for Discovery, Development and Distribution
Nature of Open Source:
Open Source, particularly the GPL V3 means no patenting or copyrighting of intellectual property, allowing physical access to data, algorithm, methods, code i.e all the constituents of the product or service to anyone who wants it , given that
1. The entity consuming the information and product has to commit to allowing the same level of openness and access to any derivative work done on the inherited work.
2. So long as condition 1 is observed there is no bar on commercialization .
Open source as a business philosophy is viable iff
a. The consumers are divided into a few who can actually tweak the product or service and others are mostly consumers but are protected from vendor lock in due to proprietary standards and code.
b. The product is hard enough to manufacture, distribute , improve while competing at the same time with proprietary alternatives with significantly large funding.
c. More eyeballs, more users, can help identify side-effects that are hard to spot in the development cycle.
Am I missing a few points. Perhaps..Do feel free to add to the list above. So now lets begin with the questions and answers about open source as a model for drug discovery-:
q1.Why would people share intellectual property e.g. interesting drug molecules if they stand a chance to build their own for-profit companies?
a1. Drug discovery is not an end in itself. What matter is making a profit or money by selling the drug molecule as a finished product either to the public or licensing it to a larger company that has the distribution network to do so.
However, most molecules that are discovered for any disease never see the light of the day as drugs. Most small companies and institutions are never able to sell their molecules for large sums to large pharma companies even after putting in enormous resources into their discovery by their modest standards. There are various economic, technical and strategic reasons for this. So no individual or entity can build investor confidence by pitching their ability at drug discovery, when the likelihood of this being licensed in a reasonable time frame for reasonable money is absent. It is only when the concept is very very novel can it pique the interest of investors. Recent example is that of a company targeting the sirtoin gene for longevity.
(Hmm, healthy longevity is another topic I have thought quite a bit about. But more of that in a later post.)
The true barrier to a molecule reaching the consumers is always the costs , risks and regulatory matter associated with clinical trials and not not drug discovery and development. Even after extensive research a drug can fail with economic consequence greater than a supernova imploding. The resultant black hole can suck in everything the company has invested in!
In fact a drug can be discovered for a known drug target(s) in 2-3 million USD, even lesser if novel computational approaches are used all along the discovery and development pathway.
Open source does not mean the company open sourcing the drug discovery data cannot make money. The IT sector with companies like RedHat has demonstrated this. As the motive of profit notive is still there, along with openness, putting faith in a for-profit open source approach, may not be a very craxy idea!
q2. Won't anyone be able to copy the molecule, particularly if the animal experiments and clinical trial results are promising?
The viral nature of the GPL permits and indeed encourages copying and tinkering, but does so provided that the fruits of derivative labour are backported to the novelty originator. Every person who makes derivatives or innovates has to pass the innovation back. In case of molecular structures there can be many potential derivatives or modifications of the original lead for optimization of certain ADMET feature or even efficacy. It is risky for an R&D institution to modify/derive a variant, use it surreptiously and sell the same. This is because clinical data or bioequivalence data has still got to be produced and approved and there can always be another variation out there that is better!
This is true even in case modifications are not made to the API structure but different formulations are produced . These formulations will still form under the gpl since they are using the API as core to the functionality, much like a software library or web service over which they apply a wrapper [formulation].
Production of a molecule should also be covered by the GPL as it is akin to building exactly the same software product when the complete specifications are well known.
There are other economic and technical entry barriers to piracy of open source molecular data. It is not an easy process to setup a government and FDA recognized manufacturing facility even when the info about a drug is known and off-patent.
Since the projects would be open source, it may get benevolent attention from government labs, NGO's and private research foundations that may contribute their data and science.
Big pharma may buy its way out of trouble by licensing rather than compete if the scale of the effort is big enough. Note that large corporations license both proprietary and open source products to hedge and manage risks.
q3. How is open source different from open science and just using open source software for an endpoint?
a3. The open source drug discovery company will discovery, develop, validate, get regulatory approval and market a Product i.e a drug . Think of it as the linux debian project or redhat! Both models work.
The science behind the discoveries of interesting molecules may or may not be open, the software used for drug design or experimental analysis may or may not be open source, but the final results would certainly be. Think about the human genome project, the hapmap project but now with commercial objectives.
So while the procedure used to arrive at the final product may not be open source the product will certainly be. It is of course desirable that processes and software be open source too but using closed source or commercial techniques would not be banned or frowned upon. Mostly information about assays and protocols is well known but may not be so in certain cases as with drug design software.
q4. How will the participating entities benefit.
Participating research institutions be it private or universities or other publicly funded institutions will be able to contribute to the science and the success of alleviating disease. The fruits of the research will not lie around in some repository for some future archaeologist to find.
Companies will not have to wait for patent expiry to start manufacturing and selling. The public will probably get access to drugs faster and cheaper. Since the information about who is producing the drug and marketing it will be public too, competition will reach and equilibrium and only those who have a clear benefit will undertake. Companies looking to profit will have early and complete access to all relevant information.
The public will see the progress of the projects in almost real time and can choose to fund projects if they choose too, just like consumers of open source software.
If there are bugs in proprietary software that are very serious for consumers and are well known, they get fixed, but it may or may not be the same for not so well known defects or bugs as there is less incentive to fix them. the situation is seriouser with drugs.
Many negative animal test results are never revealed and thus lead to needless expensive clinical trials. All along the drug disovery pathway all stakeholders will always know the pros and cons for certain features and defects e.g. toxicity. Thus some of the main fears related to disclosures will not affect this model.
This may also lead to lowering of costs of for governments whose national programs are increasing budened and who are already on a path of govt sponsored life sciences and dd research. Medicare, NHS anyone?
Governments of coutries like China, Brazil, Russia are worried about the takeover of their pharmaceutical companies by MNC's. In time there may be no competition left. Some have responded in typical fashion by starting national drug discovery programs. Russia for example. China may not be far behind as it already has a chip design program well underway.
The Indian government is keen to fund Industry-academia collaborations.
q5. Open source software companies make money by providing support rather than selling the product itself which is generally free. Would an open source dd company do something similar.
a5. Yes well it all depends on whom you trust to solve your product support issues, urgency of fixing or customization needed etc.
So selling a drug is akin to sell a software product. So what's a service? The service component is difficult to determine. But it could be using pharmacogenomics to recommend dosage or dispensing combinations or opening up of brick and mortar clinics for treatment.
q6. Is the model disruptive enough?
a6. This model for drug discovery is extremely disruptive although the NIH and EBI and other effort have already contributed substantial drug data into the public domain. Such an effort would disrupt the business as we know it. As of now it provides disincentives to eveyone concerned, the industry, consumers, government and researchers.
q7. Would there be any focus /disease areas, who would sign up?
The company would focus on signing up and getting as many labs from both the public and private research institutes, committed to the effort. Drug disovery will only be done for validated drug targets intitially in the sense that the biology for a particular disease is known to significantly affect the phenotype although it may not be known to cure it.
Data sharing would be both within disease areas as well as between them. There are several instances where a molecules useful for one indication has been found to work for something completely different. E.g. Rogaine
This would lead to synergistic effects in an open source model as drug targets are shared between several different pathologies.
Basic biology labs would be welcome, but drug discovery chemistry efforts would not start till such time as drug targets are validated.
q8. Can you elaborate on competition from large and smaller me-too viagra companies?
The critical thing about open sourcing dd is that there's less incentive for larger competitors to work on the structure since the structure and its derivatives are non-patentable.
For copycat companies, too they can only produce the molecule if they agree to an open source license and thus all formulations and derivatives info about need to be back contributed. It provides a powerful incentive to university research groups who wish to see their work have social relevance as it otherwise would never reach the public for whom these DD efforts are made.
q9. So why aren't there many such open source efforts.
The primary reason is that there is no enforceable open source license for drug discovery. GPL v2 and 3 ARE enforceable, as evinced by lawsuits against several companies that used open source code in proprietary products.
The second point is that as compared to software products there is a huge critical mass required for drug discovery. Someone has to take a lead and talk to all the different stakeholders and bring them to a common platform. Who's gonna bell the cats? Is there a Richard Stallman, Linux torvals out there to take the first critical step?
Can it be me? Well, there's a thought. But I would need Blair and Clinton's charisma, savvy and clout to make any headway!
No comments:
Post a Comment