The Chem Blog recently discussed the usability of computational chemistry software as seen through the eyes of a research chemist. A highlight:
Firstly, it's become pretty obvious that computational chemistry is a giant fucking black box. There are lots of force fields and data sets and letters followed by ** and shit and not a goddamn easy way to deal with ANY of it. Let's say I want to model a transition state, what do I do? Call a computational chemist? I'm sure they’ve got shit they want to do, too. Not only that, but after a couple decades of ripening, I'd like to think the ability to do simple computational analysis should be in the hands of the average schmuck like me.
I'll freely admit to knowing very little about computational chemistry software (a situation I hope to change). And as a synthetic organic chemist who worked in industry for several years, I tend to agree with the sentiment above. On the other hand, we had people who we paid to do these things for us. But that's a luxury relatively few can afford.
Non-experts face multiple barriers to using computational chemistry software themselves, including:
- Installation. Believe it or not, using the command line is not something the average Joe has done before. That learning curve can be quite steep, depending on the state of the system being used. Even more so for installing a compiler.
- OK, Now What? With a package installed, the questions are just beginning. What analysis do I run? How do I run it? What's all this text output for? And so on…
- Garbage In, Garbage Out. Even if we get past installation and learning to run basic routines, nagging questions persist. What does the output mean? Should I be running a different analysis? Were the parameters correct for my particular system? Did I find the right minimum?
- The Benjamins. Paying a lot of money up front for tools you know little about is not the best way to finance education. The best way to promote the real-world use of computational chemistry software is with well-supported free tools, of which there are precious few.
To be fair, there have been some positive developments in the usability of computational chemistry software - specifically the free tools. For example, Avogadro appears to be aimed at addressing many usability issues and is off to a great start.
But better software is just the beginning of what's needed. There's an entire body of knowledge hidden to the average Joe behind a wall of jargon, lore, and less than helpful documentation. For a glimpse of what might be the start of a trend, have a look at Molecular Modeling Basics, which hosts a growing collection of screencasts featuring a variety of popular computational tools.
Is this a problem worth addressing? If so, how best to do it?
Edit July 16, 2009: Added link to Molecular Modeling Basics