Biochemistry v Bioinformatics

Michael White over at Adaptive Complexity has posted an article on the growing conflict between biochemists and bioinformaticians [The problem with computational biology papers]. Michael White is a biochemist postdoc at Washington University working on cell cycle regulation in yeast so he's well-positioned to appreciate the problem.

Thanks to RPM at evolgen for finding the article. This is a problem that RPM has addressed before; you should follow his links in today's posting [Computational Work without Experimental Validation].

There are important issues here. My department has made a serious effort to develop and nurture the growing field of bioinformatics over the past ten years and we have recruited a number of excellent scientists. Nevertheless, there's still a two-solitudes problem that threatens to undermine the effort. The real bioinformaticians, and their students and postdocs, don't communicate well with the pure wet-bench biochemists. The more "traditional" biochemists still resent the fact that bioinformatics students don't get their hands dirty.

It's clear that there has to be give and take on both sides. Bioinformatics graduate students simply must learn enough biology to recognize the significance of their project. You can't get a Ph.D. in biochemistry if you don't understand what all those interaction networks really mean or you don't know how to interpret EST data.

Similary, older biochemists have to adapt to the new ways of thinking about scientific problems. There's nothing magical about working at the bench and mixing buffers. Some of the most important advances in biology are coming out of computers. This does not mean that everything in bioinformatics is valid science—far from it. A lot of what passes as bioinformatics is highly questionable, but that's also true of more traditional biochemisty. Neither side has a monopoly on truth and accuracy.