Sunday, March 21, 2010

Open source: the way forward in the search for new treatments for the infectious diseases of poverty?

Figure 1
The Open Source Discovery Foundation is based at the Council of Scientific and Industrial Research (CSIR) in Delhi.

A collaboration, called the Open Source Drug Discovery Foundation (OSDD), plans to use voluntary and open efforts to accelerate the development of affordable drugs for diseases including malaria, leishmaniasis and – first – tuberculosis.

Few if any new drugs have been created for many of these diseases, because they disproportionately affect the poor and pharmaceutical companies say they would have little hope of recouping any investments in such drugs. TB kills around 1.6 million people a year but only a handful of drugs have been developed in decades.

Zakir Thomas, project director of OSDD, says a new approach is needed: “You have to make sure the cost of discovery development is brought down to the minimal possible level by voluntary collaborations of scientists.”

OSDD, the brainchild of India’s Council of Scientific and Industrial Research (CSIR) in Delhi, is dedicated to developing drugs more cheaply by copying the model of the immensely successful open-source software movement. Here thousands of software developers from around the globe have given up their spare time to collaborate in creating software that is free to be copied, re-used or modified by anyone. (In contrast, the use of proprietary software is strictly limited to those who pay licence fees and source code is normally kept secret so any modifications, if allowed at all, are tightly controlled.)

Probably the best known open-source software is the Linux operating system but this is only one example. The mountain of software that open-source developers have created is robust enough to be used by big corporations and cutting-edge enough to have become incorporated into the latest mobile phones and laptops. But importantly, it has saved on development costs. It has been estimated that producing the open-source software we have today, using traditional means, would cost $387 billion and take 2.1 million people-years of development.

Sharing

OSDD collaborators are working on around 180 projects aimed at combating TB, using open-source principles. Members volunteer their time, share findings, papers, results, ideas and discussions online. They even share the results of wet laboratory experiments, via a so-called open-lab notebook.

Some collaborators are students, while others are high-level researchers at world-renowned institutions; some are analysing and annotating the genome of theMycobacterium tuberculosis, while others are tracking all the patents registered for TB, or finding natural products that can be tested for activity against the disease.

The idea is to get as many people as possible focussed on a problem than has ever been achieved before. And to enable this, there are strict rules. “You are welcome to use information, work further on it – you could even commercialize and make profit from it – but any improvements made on common information need to be contributed back to the community for it to work on it further,” says Thomas.

Why would so many people work on an open-source project if it is not their day job?

For one thing, they get recognition. Everybody’s contributions are attributed – via a micro-attribution system on the website. And members are able to publish papers on whatever work they have done, as long as they contribute their findings back to the OSDD to be developed further. (Some papers have been published by groups who have met via OSDD online but never face-to-face.)

Secondly, says Thomas, most of the participants are happy to satisfy their natural curiosity as scientists. Many find solving problems with a group of like-minded people immensely motivational, at least for the time being.

For some, participation is a way of getting ahead in careers. Collaborating via the portal is a chance for junior researchers – often in colleges thousands of miles away from big research centres – to be mentored by experts in their field. In theory they have access to national institutions – from India’s Institute of Genomics and Integrative Biology (IGIB) in Delhi and Central Drug Research Institute (CDRI) in Lucknow to the University of California, Berkeley in the US.

Finally, many researchers are interested from a moral standpoint. Many live in India, a country with an enormous TB burden.

Other initiatives

OSDD is just one open-source initiative in the area of biology and medicine. Others have included Human Genome Sequencing project and, more recently, the Tropical Disease Initiative (TDI).

TDI has even published a “kernel” – a term more associated with Linux than, say, leishmaniasis. But rather than lines of code, TDI’s kernel is essentially a package of predictions; 297 pairs of targets and drug ligands that could prove useful in developing drugs for around 10 different neglected diseases.

The open-source element of the project is that pairs are placed in the public domain (and cannot therefore be patented). Therefore, anyone can decide to verify the pairings experimentally to determine which might be lead compounds.

The pairs were generated by computational biologists by running completed genome sequences through complex software, called a computational pipeline.

The kernel contains 30 pairs of proteins and ligands that were identified specifically for TB, for instance. Those pairs were generated after 4000 genes of the entire M. tb genome were analysed. One piece of software looked at the sequence of genes, and predicts how the proteins will fold, their structure, and, therefore, what domain of that structure is likely to bind to drugs. A second piece of software, loaded with the compound libraries and structures of many known drugs and their ligands, is used to work out which drugs are likely to bind to these proteins.

But despite both being labelled as open source, TDI and OSDD work completely differently. Within TDI, for instance, there are only a handful of collaborators. In contrast to OSDD, there is also no impetus for any researchers, who might experimentally confirm a pairing, to share their findings with TDI. Indeed, there is nothing to stop any company from experimentally confirming a pair and then slightly tweaking the ligand, so that a modified version of the pairing can be patented.

TDI’s Marc A Marti-Renom, who is head of the structural genomics unit at Prince Felipe Research Centre in Madrid, says he has no idea who is working on the pairs he generated. That’s fine says, as long as researchers are using the information to develop new drugs. “Any group can look at these targets one by one with experimental work,” he says.

Scepticism

Certainly the analogy between open-source drug discovery and the software world seems to break down with close scrutiny say many, including Stephen Maurer, an associate Professor of Public Policy at Berkeley who has long touted open-source drug development movement to speed up R&D for neglected diseases (1).

While software developers can work free (by writing code on their PCs in their spare time), biological research is pretty impossible to do free. Although lots of drug discovery can be done on computers, wet laboratory work cannot. “Reagents cost money,” Maurer says.

That is why the Indian government has promised to pay for equipment, chemicals and contribute to the overhead of institutions that provide wet laboratory experiments and interestingly even the OSDD grants for such experiments are awarded in an open way. Applications are scrutinized by other members before money is released says OSDD’s Thomas.

The same applies to scientists’ time, it seems. Marti-Renom still has to apply for grants to carry out his work on the kernel and many of OSDD’s partner institutions allow their employees to spend some of their working day on OSDD projects. So, in fact, some open-source collaborators are getting a form of remuneration by normal salary.

Although OSDD is a great idea, the open-source model may run into difficulties when it comes to clinical trials, says Shamnad Basheer, a professor in IP Law at India’s National University of Juridical Sciences, who specializes in IP and access to medicines.

The government has pledged funding to develop lead molecules by holding reverse auctions to pick companies who can put them through clinical trials.

But Basheer is sceptical about whether companies will see any incentive to do so. After all, they will not be awarded with intellectual property rights, despite taking a risk to develop a molecule that may not work. Many of the companies who manufacture drugs in India are not experts in drug development, but generics firms whose expertise lies in creating bio-equivalent copies at wafer-thin margins, he says.

And finally, the Indian government is unlikely to be able to afford the hundreds of millions of dollars that are required to run clinical trials he says. “Where will this money come from?” he says. “[what they have already given] is nothing, a fraction of what it costs to develop a drug.”

Others are equally sceptical about the ability of the open-source model to take drugs all the way to registration. Rather, collaborations between much smaller groups are more likely, because some data is just too valuable to be shared.

That’s the view of Dr Melinda Moree, CEO of BIO Ventures for Global Health(BVGH) and ex- Director of Malaria Vaccine Initiative, PATH. Large open-source collaborations could work but only until the data they share becomes competitive, she says. Before that time, however, there are clear benefits to sharing data. She says, “Instead of wasting time and energy in all these different efforts, sharing data can be done at a pretty early stage so that other people don’t repeat what you’ve done.”

But, the incentive to share information reduces as more promising (and valuable) leads emerge she warns. “The further you get in development the more you get smaller meetings,” she says.

That probably explains the proliferation of collaborations between smaller numbers of organizations, normally on a semi-commercial, rather than open-source, basis.

Product development partnerships (PDP) for vaccines drugs and diagnostics are examples.

New approaches

Other models are being set up. BVGH, for instance, is running a so-called patent pool originally devised by GlaxoSmithKline. Together with one other company, Alnylam Pharmaceuticals, GSK has committed to donating intellectual property and know-how to a pool for 16 neglected diseases. Other manufacturers, NGOs and even university researchers can, in theory, use the pool to produce new drugs or combinations and formulations. The fee will depend on who the drugs are developed for and how much expertise is required. (Unitaid is developing another patent pool for HIV drugs.)

Collaborative Drug Discovery (CDD) is another example. It allows academics from a range of disciplines, such as pharmacology or toxicology, at different institutions to work on exactly the same molecule and build up an aggregated data file on it.

This is what normally happens in industry and is vital for drug development, says Dr Alpheus Bingham, founder and board member at CDD. (He has also foundedInnoCentive, a website where organizations can post specific drug development problems and award prizes to anyone who can solve them. It has been used on a limited basis for TB research).

In contrast, in academia, this information is normally dispersed, if it exists, across different journals. “In the academic world they are not trying to develop a drug but to answer a question, such as how well does this class of molecules bind a receptor,” he says. “But you end up with this fragmented literature which does nothing to create a candidate that gets closer and closer to being studied in humans.”

Importantly, organizations, including those interested in neglected diseases, are using CDD to work in smaller closed groups. TB Alliance, Cornell and Harvard have worked in a closed group on the CDD site, paid for by the Gates Foundation, to compare phenotypic TB screens on the same natural product libraries, for instance.

The fact is that scientists may not want to share information and their reasons may not be commercial, says Dr Barry Bunin, CEO of CDD. “Even in neglected disease area, there is still going to be some data they need to keep private because journals require it first for publication, or they may want to scoop a competitor for a grant,” he says, “Even those with the most noble of intentions have data they would like to release but cannot.”

Whether these or open-source models will succeed is unclear but, for its part, OSDD is confident that companies will be willing to manufacture and the Indian government will be able to fund later stages of drug development. And others believe that OSDD’s size, approach to mentorship and funding for wet experiments are all likely to help it succeed. As Berkeley’s Maurer puts it, if OSDD does not work, open-source drug discovery in general, may have to go back to the drawing board.

Reference

1. Maurer SM, Rai A, Sali A (2004) Finding Cures for Tropical Diseases: Is Open Source an Answer? PLoS Med 1(3): e56. doi:10.1371/journal.pmed.0010056. Available from: http://www.ncbi.nlm.nih.gov/pubmed/15630466

Comments

No comments:

Post a Comment