AMBER Archive (2000)

Subject: Re: a6 sander gb breaks on smallmol

From: Jarrod Smith (jsmith_at_chazin.structbio.vanderbilt.edu)
Date: Thu May 11 2000 - 17:42:03 CDT


"Christopher I. Bayly" wrote:
>
> I think a6 sander generalized Born (igb=1) may have a problem with small molecul
> es. When I try to minimize
> 2-bromo-4-methyl-benzoic acid (or some other small molecules) with igb=1, it die
> s after input with:
>
> Water definition for fast triangulated model:
> Resname = WAT ; Oxygen_name = O ; Hyd1_name = H1 ; Hyd2_name = H2
> Using modified Bondi radii and Tinker screening parameters
> Unable to find bonded partner for atom 14
> <end of file>
>
> The parmtop file is well-behaved; amber5-sander and amber6-sander with igb=3 min
> imize no problem to the same answer.
>
> Am I doing something wrong or is this a bug?
>

I think you've run into the same thing I ran into with my ligand
molecules. That is, the atoms need to be ordered in a particular way in
the prmtop file or the GB code doesn't work properly. The standard
amino acid and nucleic acid libraries are already ordered in this way,
but with user-defined residues like your small molecules this problem
can come up.

I solved the problem in my case by using leap to reorder the atoms such
that each hydrogen atom comes immediately after the heavy atom it is
bonded to. For methyls and methylenes, you need to list the heavy atom
followed by the 3 or 2 hydrogens bonded to it, respectively, etc.

There are probably lots of ways to do it, but the way I chose was to
write my molecule out as pdb, edit the pdb file to put the atoms in the
proper order, and then read the pdb file back in without the residue
library loaded. This re-created the residue with all the atoms in the
proper order. Then I used bondbydistance to define the bonds, and
edited the types and charges.

This might present a problem if you had a lot of these to do. There may
be an easier way. Maybe you can edit the .off file directly. I'm not
sure if the bonds in the .off file are dependent on the atom numbers or
the names only.

Hope that helps.

-- 
Jarrod A. Smith, Ph.D.
Assistant Director, Center for Structural Biology
Molecular Graphics and Computation
Vanderbilt University

jsmith_at_structbio.vanderbilt.edu