[gmx-users] pdb2gmx

Justin Lemkul jalemkul at vt.edu
Wed Oct 29 18:19:20 CET 2014

On 10/29/14 12:55 PM, Eric Smoll wrote:
> Hi Mark,
> Thanks for your help. I am looking for a deeper understanding "residue
> names," "atom names," and how they are used to properly form input for
> grompp.
> By experimentation, it appears the pdb2gmx requires unique "atom names"
> within each residue of the .rtp file to properly process .gro/.pdb files.
> Thus, every "atom name" of a residue in the .gro/.pdb file must be unique.

This is correct.  There's no other way to process the input file otherwise, 
because neither atom order nor atom numbering can be assumed.  The only thing 
that should be reasonably foolproof is atom naming.  Internally, pdb2gmx then 
maps the names within the residues to numbers before writing out the processed 
coordinate file and topology.

> It this not made clear in the .rtp section of the manual. It also appears
> that the bond directive is required, not optional as shown in the example
> in the .rtp section of the manual.

Strictly speaking, [bonds] is not a required directive.  One can specify 
single-atom .rtp entries (e.g. ions) that will not have bonds.  So making bonds 
mandatory breaks any case that may have, for instance, coordinated ions.  So a 
[bonds] directive is required only if the residue has bonds.  That's certainly 
the majority of cases, but it would be wrong to say the directive is mandatory.

> However, if I build a system using an .itp file and the [ molecules ]
> directive, the requirements are more flexible. Again, I observe that while
> the "atoms names" of the gro and .itp files must match within a given
> residue, they need not be unique. This is not made clear in the manual. It
> seems (I hope) that matching is accomplished by the "nr" field of the
> .top/,itp file and the "atom number" field of the .gro/.pdb file.

grompp does check for atom name matches as a sanity check.  It will complain if 
there are mismatches, which guard against a user preparing everything manually 
and then hosing the simulation by having a different order.  Indeed, there is no 
requirement that atom names be unique here, but order is essential since all 
remaining interactions (bonds, angles, etc) are based on atom number.

The atom number field is not the deciding factor.  It's matching the order of 
the names in coordinates vs. topology.  See the "check_atom_names()" function in 
grompp.c in the source code.



Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 629
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441


More information about the gromacs.org_gmx-users mailing list