This was what Klaus Schulten heard from an experimental colleague after he had told him what he’d found and it neatly highlights one the key advantages of molecular simulation; you can see in extraordinary detail what is happening or, as Klaus put it in his National Lecture last night,
It doesn’t only agree with the experiment, it tells you more.
Implicit in that statement is that the simulation must first agree with experiment, which is pretty obvious, but still worth saying! The best way of achieving this, in my experience, is that you work closely with experimentalists, preferably from the beginning of a project. Klaus did not explicitly spell this out it but it was clear from his lecture that this is what he always tries to do.
He also showed several very nice examples from ankyrin repeats to cadherins to aquaporins where simulations from his group have made predictions that were subsequently confirmed by experiment. Predictions of this kind are, I think, the possible the best demonstration of the utility and power of computer simulation in molecular biophysics. Another key theme was choosing the appropriate level of description for any simulation, or as Klaus put it,
Chemical detail is important.
This is true; we must resist the urge to discard detail to make our simulations simpler and faster at the expense of biological accuracy and instead chose the most appropriate description for the system at hand.
Given the audience was mostly experimentalists it was striking how long Klaus spent talking about the tools i.e. the software, or to be specific the molecular dynamics (NAMD) and visualisation (VMD) codes that his group have developed and now supported for around twenty years. But then again these are important and both NAMD and VMD are very widely used so why not talk about them?
What is unusual is that both codes have a full-time developer which allows them to make regular releases, have good documentation and also nurture a community of academic contributors. I think the question this raises should be why don’t more academic codes have this degree of support? And, in tandem with that, why don’t we as a community use more software engineering tools, as encouraged by Software Carpentry and institutes like the Software Sustainability Institute in the UK.
That Klaus ended with an advertisement for a training course is especially poignant for me; I attended the first NAMD summer school back in 2003 and this really kick-started my PhD (I was at the end of my first year). As an aside I enjoyed the anecdotes and stories during the lecture, like the one I started with, and so I thought I’d end with one of my own. During the summer school in 2003 I have a vivid memory of Klaus addressing us all and booming out “if you use molecular dynamics to calculate free energies you are braindead!”. So when I saw those free energy profiles last night I smiled. I’m glad that even distinguished scientists can change their minds.