Can Open Source Work in Chemistry?
OK, I’ve been on a bit of an open-access/open source kick with posts recently. But I’m going to post one more and talk about something else next time.
I heard quite a few people at the ACS meeting expounding on the opinion that open source won’t work in the chemical field. The two most common arguments were:
- The most widely-used chemical software is proprietary. If open source is so good, why don’t more chemists use it? Particularly, why don’t the more computer-savvy theorists and computational chemists use open source software?
- There aren’t many chemists who write code, much less write good code. Who will develop open source chemistry applications?
So here’s my $0.02 on the matter…
How about a case study? What if there were a computational chemistry program that:
- Started with an academic codebase.
- Grew through countless contributions from academic, government, and industrial researchers. Easily hundreds of top-notch research codes have been contributed over the years, including the initial codebase.
- Distributes source code worldwide and supports a wide variety of computing platforms.
- Runs a wide variety of computational chemistry models, far beyond its original purpose.
- Is used by a large number of international groups, including academic, industrial, and government labs.
- Has immense name recognition and market share.
Sounds like a perfect example of open source software succeeding in the chemical market, doesn’t it? I’m not going to give names, but in fact by a quirk of history, in my opinion, this particular package is proprietary, is run by a company with questionable business practices (banning various academic groups (customers), for example), and strange licensing restrictions (banning public discussion of performance).
If you work in computational chemistry, you probably know which program I’m discussing. I bring it up, however, to point out that it’s a case study for how open source could work.
It addresses the question of whether there are enough chemists with programming skills — there certainly are enough to maintain a variety of proprietary and open source applications right now and clearly current and future graduate students have greater and greater familiarity with computers and computer science skills.
As to the question of the use of open source chemical applications, I think that’s a matter of time.
4 Responses »
Comments
Subscribe to comments on this post via RSS or TrackBack
Peter Murray-Rust — 3 years, 8 months ago.
I gave a presentation at the ACS on the values of OpenAccess in chemistry (the idea that publicly funded chemical research should be publicly available). It also extends to Open Source, Open data and Open Standards. See:
http://wwmm.ch.cam.ac.uk/presentations/acs2005/index.html
[You will need SVG (downalod from http://www.adobe.com/svg + IE to see the pictures]
BTW Geoff has done great work in continuing to develop OpenBabel - an important Open chemistry program
Geoff — 3 years, 7 months ago.
Peter flatters me. I’m hardly the only contributor to Open Babel — he contributed a large chunk of CML code as well as some other important pieces.
As I remarked to Peter during the ACS conference, I think one advantage of open access is that it’s “sticky.” In the terminology used by Malcolm Gladwell in his book Tipping Point open resources often snowball because it’s hard to reverse the trend. And much like Peter, I don’t think the issue is necessarily finances but access to the data and research. Why cannot the public, educators, or other researchers obtain my data sets for their uses? Sometimes they can, but not always?
In the context of programming and coding (i.e., this post) I think it follows — why can’t I obtain the code or model from a journal to reproduce or build on another research article?
Noel O'Boyle — 3 years ago.
I am currently counting up the number of open-source computational chemistry packages. My grand total so far is three: PyQuante, MPQC and ABINIT. Can you think of any more?
Geoff — 3 years ago.
Noel, since I’m willing to include molecular mechanics / dynamics codes and DFT codes from physics groups (you include ABINIT), there’s a lot more.
DFT++, Gromacs, YaehMOP, PWSCF, Moscito, Moldy, CAMPOS, MMTK, Spinor, octopus, CP2K, Dacapo, PEtot, Socorro.
(I’m not sure if that’s it. These are codes, where I’ve checked the license.)
There’s actually a pretty good list at FSAtom.org. (I believe this dates from a conference at CECAM.)
There are also several quasi-open source chemistry packages, i.e., free redistribution from the original group, source code available. Perhaps they might move to a more open license, maybe not.