Alan, Last I checked, MSMS was saddled with a number of constraints on its usage and redistribution. Though source code can be obtained under certain conditions, MSMS doesn't meet the minimal "open-source" software requirements that would make it eligible for inclusion into PyMOL. I talked with Michel Sanner about this a couple years ago, but he didn't think there was much that could be done. (DISCLAIMER: I haven't double-checked this since -- so things may have changed since without my knowledge).
The other problem with MSMS is that it bombs with large or convoluted structures on machines which don't have enough stack space. Plus, I believe it is written in C++. PyMOL current only depends on C and Python, and am I loathe to introduce a new language dependency without a compelling reason. Though it is a matter of personal preference, some of us actually prefer the appearangce of surfaces created by PyMOL and Grasp over those created by MSMS or Insight, which have characteristic "streaks". No question PyMOL is slower to calculate surfaces though. The algorithm was created in a single weekend and is very inefficient. I've spent considerable time trying to optimize it since, but have only been able to boost performance by about 10-30%. I think the order of this naive, brute-force algorithm is just too high. If someone can find a better open-source surface calculation code, I'd be happy to include it in PyMOL as an option. The only catches: it must be written in C, and it must be available under a BSD-style open-source license (free for commercial use, with no limits on redistribution). Anyway, approximate partial surface areas can be obtained in PyMOL using the dot representation, but that isn't yet documented... Cheers, Warren On Mon, 10 Feb 2003, Alan Wilter Sousa da Silva wrote: > Dear Dr. Warren, have you consider using MSMS to generate surface >in Pymol, as VMD already does? > Since MSMS can generate a list too, by using another programs, I >could calculate properties to each point in the mesh. So, I wonder, >applying a colour method linked to such properties... well you can >imagine.
