[MMTK] Atom ordering and some other problems

Jaroslaw Kalinowski jak at biogeo.uw.edu.pl
Mon Jul 16 16:25:30 UTC 2007


I have yet another set of questions about atom ordering. I know that it is
recommended to use atom names only but unfortunately the concept of atom
ordering is useful. First there are external programs that identify atoms
by number rather then by name (like most quantum mechanical codes or
programs to  calculate Poisson-Boltzman electrostatics), second it is
sometimes easier to use plain arrays rather than objects. For example
ParticleProperties are fine for representing properties defined for each
atom in a universe, but how could one represent data for selected objects
in the universe, or how could one represent two-atom properties like bond
orders defined for each pair? Arrays seem to be a choice, but for that one
has to know ordering of atoms. I would appreciate being able not only to
get a mask for an object but also get indices that would allow to use
take/put functions. Now the questions:

1. Ordering given by u.atomList() is not the same as ordering of
configuration array, is it?
2. If not than how one can iterate over atoms in the same order as is used
for arrays?
3. Is it possible to get atom number in pdb file for peptide chains after
they have been created using PDBPeptideChain.createPeptideChain?
4. Does pickling and unpickling preserve atomList ordering and/or array
5. Does using copy.deepcopy do?
6. Does saving using PDBOutputFile and than reading using PDBConfiguration
7. And can one know in advance what would be atom ordering in the pdb
8. What is the atom ordering in NetCDF trajectories?
9. Can one assume that each residue has a continuous slice in
configuration array?
10. If so, than is the atom ordering in such slice identical for each
residue of a given type?


Some more problems.

Is it possible to store additional data in a trajectory? I have a QM
forcefield that produces atomic charges, bond orders etc. Is it possible
to have my function called to provide data for each trajectory frame?

I would like to try doing QM/MM calculations. It seems that it is not
possible to apply Amber forcefield to the part of a system, so I was
thinking about cloning MM part and adding QM part to a copy. Than I could
just write a forcefield that would just copy energy and gradients from MM
universe and add my QM terms. Do you think it is a good approach?

It is not possible to change protonation state after PeptideChain has been
created, is it?

I am sorry for rather lengthy list of problems, and thank you in advance
for answers,

Jaroslaw Kalinowski

More information about the mmtk mailing list