The Hoard

Science is built on the daring and unconventional ideas of a few going against the hive mentality of the pack. Things that seem like strange artifacts in the data emerge as cryptic signposts pointing to bold new ideas. The bizarre non-Newtonian behavior of light led to relativity, and the equally bizarre downregulation of supposedly overexpressed genes in petunias eventually became RNAi.

But in the era of big science, how do we pick what anomalies are worth following up on? As science becomes more and more dependent upon technology to progress, any line of inquiry becomes expensive both in dollars and man hours. Within this paradigm, is independent investigation possible?

I would argue that in many ways it is not, but our current model still does little to promote the large scale collaboration necessary for modern research. There is still a big emphasis (in biology at least) on first author publications rather than roles in large collaborative studies. Collaborative openness, while often touted as a cornerstone of an institution, often does not make careers. We like to put a name on a given discovery, or maybe two names. We award Nobel Prizes to a handful of researchers, not a collaborative team.


That's not to say that we're not trying. Take The Cancer Genome Atlas (TCGA). This massive effort by six major research centers will work to sequence tumors and matched normal tissue from the same individual across dozens, then hundreds, then thousands of individuals. Eventually it will trace the contours of the cancer genome at extremely high resolution by using very powerful (but expensive) sequencing technology on a massive scale. This is the sort of google maps of the cancer genome. This atlas will be available to everyone, publicly, at no cost to the users.

It will be interesting to see if this new paradigm works. As great as these new, massive, collaborative databases are I think we have one risk, which is a sort of inverse tragedy of the commons. If the data is public and everyone has access to it, no one will have a vested interest in following up on that data. Publishers just aren't as impressed with computational follow up as they are with studies that generate new data. Furthermore we seem to have very little interest, in our modern scientific society, in 'negative' results. Much of what these databases will do, I think, is wash away the apparent significance of correlative studies done with our scientific eye trained on just a few pathways. If we look at two genes and see that they go up and down together we call them a signaling pathway. But if we look at 2000 genes and see them go up and down together, do we still come to the same conclusion?

It's a rocky road we embark upon.