A different take...

Erik's article is thoughtful and thought-provoking - exactly what PLoS can do so well. Having shared an office with Wil Rall just as he was working out the importance of dendritic spines, I completely agree that he changed the face of computational neuroscience. On the other hand Erik's article misses many of the schools of quantitative biology that were important 30 years before the phrase "systems biology" came into the scientific lexicon. And some of these schools, contrary to Erik's assertion, had enormous influence on the biology they studied. Two come immediately to mind.

In the 1950s through 1980s Mones Berman worked at the NIH first in Zim Hearon's Mathematical Research Branch (just across the hall from Wil Rall) and later ran his own show in the Laboratory of Mathematical Biology in the NCI. Berman was an engineer, and a pioneer in the application of compartmental analysis to biological systems. But he published in biological journals. And he had impact. The methods he developed and several generations of his students still dominate the field of lipoprotein metabolism. The meeting he initiated as a satellite of the American Heart Association, just celebrated its 40th anniversary.

In that same era there was Arthur Guyton, a physiologist who would have laughed at the notion that systems biology is new. Arthur once told me that his secret was he never told a study section about his models. Instead he used those models to design great experiments, and then he told the study section about those experiments. But there are hundreds of cardiovascular scientists, both basic and clinical, whose view of their discipline was profoundly shaped by Arthur's models and his books. He too published in biological journals.

To me the question is NOT why computational neuroscience and systems biology are so separate. The question is: Why is computational biology so separate from biology? My guess is that the gulf between them is caused not by arcane mathematical methods or by biologists self-selection away from anything quantitative, but rather by the apparently well-entrenched desire of computational biologists to publish in places other than the journals where their experimental colleagues are publishing. And vice versa.

It would not surprise me if academic departmental boundaries also played a part. If your training is in mathematics and you are surrounded by mathematicians, it is natural to want to succeed by publishing new mathematics, not new biology. Similarly for physics.

The young people I meet at ICSB are not very different from those I meet at the American Society for Cell biology. By and large, they all believe that biomedical science must become more quantitative. And if you look carefully and are willing to take the long view, there are more and more modeling papers being published in Science, Nature and Cell. The same is true of the more specialized biomedical journals.

These articles get read by experimental biologists because they cannot afford to ignore them. They get read because they advance our understanding of some important biological system; in short, they answer outstanding biological questions. And often they are written by teams. Teams from across the spectrum from theory to experiment. A wonderful recent example is the paper from the Theriot and Mogilner labs: Nature (2008) 453:475-80.

Teams are very hard to assemble in academic biology despite Elias Zerhouni's hard work in reshaping NIH policy. Because this is so difficult, great teams of computational and experimental biologists come together only by chance.

Consequently, what I've been telling the young people at both ICSB and ASCB is they should both be aiming to publish in the same journals and they should show up at one another's meetings. Very few of them have ever even considered this possibility.

This is the gap that needs bridging.