On Mon, Jul 29, 2002 at 01:12:40PM -0400, Nathaniel Echols wrote: > How hard would it be to incorporate the DSSP algorithm? I was looking at > the source for RasMol, which has a complete implementation. My feeling is > that this could be ported without too much misery. However, RasMol's > license terms are different, and I suspect even a complete translation > into Python with the ChemPy API would still be considered a "derivative > work", meaning it wouldn't be able to be part of the main distribution, > and meaning someone would still have to code DSSP from scratch. (I have > utterly no idea how it works, but RasMol's code seems clear enough despite > lack of comments)
I am not a lawyer, just a scientist who happens to be interested in these sorts of 'intellectual property' problems, so weigh the following accordingly: Copyright and the idea of derivative works applies to very specific expressions of an idea. So, unless the DSSP algorithym is patented, a complete re-expression in a language like Python (instead of C or C++ or whatever RasMol is written in) could probably be made different enough so as to be an original work, distinct from RasMol's implementation. This is what the the Gnu project did with Unix command-line utilities for example--rewrote them, added features, and so forth. The best way to avoid "derivative work" concern is to code in a "clean room" fashion, without reference to someone else's code. In principle, this could be done working from the original paper(s) that describe an algorithm. Not that I'm volunteering ;-) Just that I would hate to see someone who was willing and interested in implementing secondary structure code for PyMol to be frightened off the job if, indeed, it could be done in a way that respects others' rights. --Joe
