Thursday, February 17, 2011

Roads That Really Need Taking

There is a provocative Perspective piece in a recent Nature titled "Too many roads not taken". In it, the authors summarize some more extensive work (posted on Arxiv) and others showing that in several major fields of protein research, the biological community seems to be stuck on a few old favorites, with insufficient attention to other key players.

In the Nature piece, the focus is on protein kinases and nuclear hormone receptors, both key druggable classes. In the kinase area, they have a graph (the Arxiv site leads to the actual data) showing that the graph of attention (measured in bibliographic citations) given to kinases prior to 2003 is generally predictive of the attention given them in 2009; only a handful of kinases with little attention in the first time period are hot areas now. Of course, if the rest of the kinome were boring this would be justifiable, but they point to several papers which indicate that many of the unstudied kinases (for example, this series of shRNA experiments)

A very striking conclusion is underlined by their plot of NHR citations by gene, which is expanded upon significantly in the Arxiv preprint. Their contention is that the availability of chemical probes drives analysis in this family, and in their rank-ordered histogram of activity the proteins with chemical ligands are all at the top of the list. Granted, at the breakpoint the bars are nearly equally tiny on either side of the divide, but it does suggest a strong correlation. Easy to build universal approaches such as RNAi are very powerful, but especially for any kind of in vivo study are suboptimal -- not only are they hard to deliver, but utterly wiping out a protein is not a subtle way to probe a system. The Arxiv piece goes into more detail and posits that you really need good compounds, ones with good bioavailability and such. In other words, drug-like compounds.

So, they plead that we need to invest more in getting such compounds into the research community. Some of this could take the form of funding screening efforts in the public sphere, which like sequencing the human genome isn't the sexiest project from a scientific perspective -- just like the foundation is never the sexiest part of a great building. But foundations matter!

An alternative, suggested by the Arxiv paper, is that compounds may exist in the patent literature but are not widely commercially available. So, a public effort might focus on mining the patent literature, synthesizing various claimed compounds and verifying their activity. Of course, some (many?) compounds in the patents might not stand up to scrutiny. Another alternative would be to attempt to devise incentives for industry to release some of the inhibitors they have in house; obviously this would be a challenge as such compounds might prove to be the starting points for valuable drugs -- but of course they never will until someone proves the underlying kinases valueable. Some companies have generated unbelievable databases on kinase inhibitors (for example, this recent study) which have the potential to fill in the missing spots on the kinase tool compound grid.

I can't help thinking of the new translational medicine institute going in at the NIH. Trying to move research findings into the clinic to benefit patients is hard to argue against, but perhaps the NIH needs to also pour some resources into the humble task of finding basic tools.


Rick said...

Not to be nit-picking, but usually the site is referred to as arXiv


I believe is the reason behind the name. Although I am hardly from the Physic community so do not know for certain.

Keith Robison said...

Thanks! I was raised to be a stickler for spelling & typography so I'm a bit embarrassed to mess that up.