Hello Egon and All! Thanks for the response and the ideas. Well, yes I also think a new MMFF94 atomType mapping would be useful. Currently the 3d builder doesnt work properly, bacause the CDK atomtype to MMFF94 atom type mapping is not correct. This is mainly because now those atom types are assigned through HOSECODEs and Java regex matchings.
One possible way to go on is to step-by-step test the ForceFieldConfigurator assigner method for lots of molecules and as a new wrong atom type matching occurs, make a patch for it in the regex matcher. (In this case we assume the HOSECODEGenerator class works properly.) This is what I started and even in the first 4-5 drug molecules I found 3-4 bad mmff94 atom types matches. They are ready for bug reports and I will "patch" them unless we follow the second way. The second way is what Egon suggested, that write an MMFF94AtomType matcher. That would be better and we dont have to use the hosecode trick then. So I think that would be good to set up a GoogleDoc for the atom type pairs between CDK and MMFF94. The last thing is that I suggest we should omit the MM2 case. MMFF94 has more atom types and most of the users use MMFF today, so it think it might be reasonable to dump MM2 from CDK and later introduce other FFs that can be used for e.g biomolecules. That would extend ModelBuilder3D. ------------------------------------------------------------------------------ For Developers, A Lot Can Happen In A Second. Boundary is the first to Know...and Tell You. Monitor Your Applications in Ultra-Fine Resolution. Try it FREE! http://p.sf.net/sfu/Boundary-d2dvs2 _______________________________________________ Cdk-user mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/cdk-user
