Le mardi 31 janvier 2006 à 11:05 -0500, Geoffrey Hutchison a écrit : > On Jan 26, 2006, at 4:23 AM, Jean Bréfort wrote: > > >>> 2. Generate "reasonable" 3D structures from 2D. > >>> 10. Generate multiple "reasonable" conformations of a structure > >> > >> All of these are really going to require access to a force field or > >> MOPAC or somesuch. Like Michael, I've thought about accessing > >> libghemical where available. I think this may make some sense, but > >> maybe someone will contribute a force field. (We have donated code > >> for a tripos-like field, but the included parameters weren't > >> copyleft.) > > > > Hmm, libghemical needs openbabel. Circular dependencies must be > > avoided. > > May be libmopac would be enough ? > > No, I don't think libmopac is a good idea here. Most people, I think > want some sort of force-field driven 3D structures and conformations > FAST. > > I don't know the best way to handle the potential circular > dependency, but here's one thought: > * Open Babel checks for libghemical to incorporate the force field, > conformer generation, etc. > * Libghemical removes the dependency on Open Babel (the current code > actually doesn't use libopenbabel in libghemical) > * Ghemical (the program) still relies on libghemical *and* > libopenbabel where available. Seems correct. Another possibility would be that libghemical adds a module to libopenbabel. Imagine we have an OB2Dto3DConverter, libghemical might implement its derived class and register it (and other programs might do the same). Cheers, Jean