There was a post on In the Pipeline a little while ago that has been rattling around in the back of my mind, about open source drug discovery. That article referenced a column by Jackie Hunter that used a lot of management speak but did not really say anything about how something was going to be done. It left me with a nagging question: what does it mean to have an open source drug discovery effort?
Open source to me at least means a lack of stake-holders, that anyone and everyone who cares to can contribute to the overall project. With something like Linux, an individual might work up a new application or improve upon an old one. People using the operating system can make use of that new application or not. It is very fluid and modular. Another example is the way mathematicians are tackling proofs, as is talked about a bit in this New Scientist article.
There are a couple of reasons these don’t really work for drug discovery. The first is that while there are many components to a drug discovery project, they are not as such modular – they need to work together effectively. You might be able to re-arrange the order according to preference or availability, but ultimately, you can’t just run random assays without controls; the data generated would be huge, expansive and a total mess. The other point is that writing software or thinking about math proofs does not consume anything except time, assuming for the moment that the user has a computer to work on. You might make an argument for coffee as well. But in contrast, drug discovery uses a lot of equipment, chemicals, just stuff. Someone has to pay for it and people don’t like to pay for something when they don’t get something in return.
That thought led me to work by charitable foundations, such as the Gates Foundation amongst others. They have money and a will to tackle a pressing medical need. Thus they form the hub of such an open source project, coordinating the effort. They have a network of collaborators, whom they can call upon if they need a particular assay or experiment run. But it is not really open source. It cannot be totally open because they cannot have people randomly contributing to the project.
I think what Jackie Hunter and others are really talking about when they say “open source” is a more fragmented and specialized pharmaceutical services industry. The big pharmaceutical companies pare back their internal departments to a minimum but when they need something, they can still get it from an outside source. Not open source but out-sourced.
The only way chemistry can be done via an open source collaboration is either through idea generation or on a small scale. For the latter, I am thinking of testing a reaction procedure – usually one the originator of the request has some doubts about. For the former, I generally find that people are very happy to contribute their thoughts and ideas. Sometimes the harder choice is to sort through the good from the bad. Derek Lowe pointed us to this effort from the University of Sydney. A few commenters to that post expressed doubts that such a crowd-sourcing approach could work. Is it an efficient way to approach the problem? Could they end up spending a lot longer working through possible solutions when an experienced expert could solve it in a fraction of the time? Perhaps. I like to think that we might contribute our ideas during a slow time at work and make some little difference.