It would be interesting to have benchmarks testing the
force field based geometry optimization in the CDK.
Specifically, consider a a number of sets of molecules
- small drug like, polycyclic rings, proteins etc.
Then start with initial structures and run
optimizations using the CDK as well as other codes such
as semi empirical (PM3 in mopac) as well as other force
field based codes. The quality could be measured by the
RMSD for structures optimzied using CDK and other codes.
In the case of small molecules the reference structure
would be PM3 or ab initio based optimizations
(Gaussian?). For protiens the reference structure would
probably be obtained from semi-empirical or MD
Log in to post a comment.