[gmx-users] GROMACS state-of-the-art setup (was Re: Intermolecular Distance Restraints)

David spoel at xray.bmc.uu.se
Sun Aug 15 20:42:23 CEST 2004

On Sun, 2004-08-15 at 19:32, Warren DeLano wrote:
> Thanks for the help...pdb2gmx -merge works like a champ.
> As far PyMOL goes, I'd like to get a better handle on what the state of the
> art is with respect to setting up GROMACS simulations.  I'm hoping someone
> can answer the following questions:
> - Is there currently a way to restrain sub-portions of a system using a
> conceptual selection language (ala XPLOR/CNS/PyMOL), rather than atom
> identifiers?

> For example, in the past I've needed to do stuff like the following.
> select ligand, chain A
> select protein, chain B
> restrain_position (all)
> unrestrain_position (ligand)
> unrestrain_position (protein within 10 of ligand)
> restrain_position (protein///143-144/)
> restrain_distance (ligand///5/N), (protein///123/O), 3.2
> restrain_distance (ligand///6/O), (protein///122/N), 3.2
> in order to set up models where certain interactions or conformations are
> assumed/enforced from the get-go.  
> The above isn't all that far from being doable with PyMOL by using Python to
> simple read, edit, and write posre.itp and disre.itp files, but is there
> already some other tool which can do this efficiently for GROMACS?
Yes, you can do such kind of atom selections using the make_ndx program
with simple logical operations such as AND and OR (&, | respectively),
probably except for the distance related bit. Whether that's a better
solution than doing it from with pymol I cannot judge.

> - Is there an easy way to avoid renumbering of residues and chain
> identifiers?  or a way to correct the final output from GROMACS so that it
> matches the naming used in the input PDB?
Naming probably not (you know why...). You can however generate an 
index file with pdb2gmx -n that will give you back the atoms in the 
original order (which is, as you know, quasi-random in pdb files). Upon reading
a pdb file we store the original numbering etc. but this does get lost 
when doing simulations.

> Setup, analysis, and sharing of results is harder if you're constantly
> having to renumber systems, and especially when you're using fixed
> identifiers in atom selections as in the above examples.  PyMOL could
> probably handle the remapping automatically stored the original identifiers
> in Python dicts, but perhaps there is already a simple solution to this
> problem?
> - Is there any GROMACS tool for automatically assigning semi-reasonable
> charges and types to arbitrary organic molecules?  

> I'm not talking about QM-quality parmeterizations, but rather the quick and
> dirty (yes, mostly bogus) stuff that industrial computational chemists need
> to do on a routine basis in order to rapidly generate crude models of
> protein/ligand complexes.  PyMOL has a chemical substructure pattern
> recognition module that could be useful for this...

David van der Spoel, PhD, Assoc. Prof., Molecular Biophysics group,
Dept. of Cell and Molecular Biology, Uppsala University.
Husargatan 3, Box 596,  	75124 Uppsala, Sweden
phone:	46 18 471 4205		fax: 46 18 511 755
spoel at xray.bmc.uu.se	spoel at gromacs.org   http://xray.bmc.uu.se/~spoel

More information about the gromacs.org_gmx-users mailing list