Common Knowledge

The Tropical Disease Initiative has released a "kernel" for open source drug discovery. It's been published in both Nature Biotechnology (ugh, subscription required) and in PLoS Neglected and Tropical Diseases (yay, open access fulltext under CC-BY).

I am not steeped enough in the reality of drug discovery to make believable statements about how much this means for those actually on the ground looking for cures. I do know that drug targets are only the first step on a really long road to drugs in patients, and Derek Lowe has written a very informative post on this topic over at In the Pipeline. Suffice to say there's a lot between the gene and the cure.

What I do know is that this first step of drug targets is critical enough that companies have risen high and made a lot of money by locking them up with patents and trade secrets. At the risk of overloading Derek's blog, his post on "ten years after: the genomics frenzy" is essential reading. There was a mad land grab going on for a long time, resulting in statistically shocking amounts of the genome being patented (although the impact hasn't been as bad as it could have been, thanks to better regulations at the USPTO and recent court decisions).

This kernel (which is in reality more like a database than the Linux kernel) represents a shot across the bow of those who would argue that disease biology should be a competitive space. Like the genome itself, it's becoming more and more evident that collaborative behavior and pre-competitive action is the best management scheme for disease biology. Property behavior doesn't lend itself to innovation in a space of such enormous and disparate complex knowledge.

I'm thrilled that the data underneath the kernel is itself available in the public domain under the Science Commons data protocol. This is a wonderful step forward in the voluntary construction of the public domain - had this been done at the NCBI it'd have been public as a default step, but the choices of the TDI have ensured that the data can be effortlessly and endlessly remixed into other public domain data sources. They understood that it's about interoperability, not control, lowering resistance to the public's right to share, not compelling behavior on users. And in so doing they've done a lovely job of creating a ton of prior art that itself should help avoid dumb patents skating through the USPTO.