[gmx-users] First release, translation supporting PDB<->(AMBER, GROMACS)
mathog at caltech.edu
Tue Mar 21 00:06:18 CET 2006
I've been working on table driven translation between PDB
and AMBER/GROMACS "PDB" files. That is, getting things that are
nominally in PDB format into, through, and out of the GROMACS package
without having to edit PDB files. Especially structures that used
DNA and Protein, and so had to use the AMBER FF files.
For instance in a "PDB" file the same atom might be written:
(PDB) T:5CM, (GROMACS) THY:C7, or (AMBER) DC5:5CM.
The flip side of my changes is that if you encounter new
residues or atoms the translation tables may need to be edited,
but that can be done once, versus every time for the PDB files.
The pieces needed to implement these changes are here:
Unpack that from one level above "gromacs-3.3" and it should put
program files where they need to go. (If you have local changes don't
do this, it may stomp on your changes!) Then follow the instructions
in AAAREADME_CIT.TXT, which will be in the top level. In short,
move the two name translation tables (for residues and atoms) to
"top", set up AMBER and FF.DAT in "top" as well, and then use
./configure, make, and make install normally and the changes should go in.
Programs that are changed are:
pdb2gmx: on my test cases did a good job of PDB -> GROMACS
(if no DNA) or PDB -> AMBER (even with DNA). If protons have
been added, for instance with "reduce", -ignh is generally needed
to go to GROMACS but not to AMBER.
trjconv: with -dump N -ff AMBER (or -ff GROMACS, the default) now
exports a PDB file that most programs can read. Generally you
will want to export the entire system and use -nowet to remove
the waters. If -wet (the default) is used a zillion waters
go into the PDB file as ATOM when they should be HETATM. Eventually
the program could guess the -ff, but for now you must tell it
which one you used with pdb2gmx.
Note that trjconv still uses the default GROMACS classification
code, as does ngmx and the rest of the GROMACS package. Consequently
AMBER residues will typically end up in their own groups since
these existing routines have no idea what DC5,HIE, etc. are. But
trjconv only lets you select a single group to export. That is why
you should select system (group 0) and remove the waters (generally)
I do not guarantee that any of this works correctly. It seems to
for my test cases, but treat this as an alpha release and proceed
at your own risk.
mathog at caltech.edu
Manager, Sequence Analysis Facility, Biology Division, Caltech
More information about the gromacs.org_gmx-users