From: Peter Freddolino (petefred_at_ks.uiuc.edu)
Date: Mon Mar 20 2006 - 16:50:03 CST

Hi Brian,
this turns out to actually be a fairly common problem that people
encounter with some topology files. Until version 31 of the charmm
topology files, charmm and the PDB maintained a different nomenclature
for the leucine delta carbons; as a result, hydrogens would be placed in
the wrong place. There is really no standard for how this assignment is
done, so depending on where you got your pdb file from you'd need
different topology files or aliases to have things work correctly; a lot
of people have used different fixes for this. Unfortunately, I didn't
set up the right aliases in autopsf; this will be fixed in the next
version. In general, if you're dealing with a topology file that doesn't
give you the right leucine HG position, you can fix it by either using
the aliases
pdbalias atom CD1 CD2
pdbalias atom CD2 CD1

or by deleting the line
IC CD1 CD2 *CG HG 1.5361 110.2600 -116.6300 108.0200 1.1168
from the LEU entry of your topology file, which will cause psfgen to
guess the position of this atom (which it will correctly place).

I should stress that the former of those two solutions should work right
for most RCSB-derived PDBs but is incompatible with the charmm
nomenclature; the latter solution should work for all cases (although
the H-C bond distance will be slightly off because you're not using the
distance provided in the IC entry; this shouldn't be much of a problem
because it will just default to a 1.0 angstrom bond instead of 1.12).

To help people notice problems of this type, psfgen will give warnings
and automatically guess coordinates of this type where the coordinate
generation results in bad contacts. This really seems to fall into that
category of problems with computers doing exactly what their input tells
them to.

I hope this is helpful please let me know if you have any further questions.
Best,
Peter

Brian Bennion wrote:

> Hi Peter
>
> After I sent the first email regarding autopsf, (apologies to those
> email clients that choked on a 1.4M attachment), I ran the same pdb
> file through my standalone psfgen that came with vmd1.8.3 and still
> saw the same problem.
>
>
> brian
>
>
> On Sun, 19 Mar 2006, Peter Freddolino wrote:
>
>> Hi Brian,
>> could you please send me the topology file and script you used to run
>> the psfgen standalone on this structure? I am able to reproduce this
>> error, so if you send me the rest of that data I should be able to
>> fix it.
>> Many thanks,
>> Peter
>>
>> Brian Bennion wrote:
>>
>>>
>>> Hello,
>>>
>>> I have been testing the new features in 1.8.4b11 (sorry I know I am
>>> one behind) for windows and noticed that placement of hydrogens on
>>> the branching Carbon of Leucine sidechains is placed parallel to the
>>> axis of the Cbeta-Cgamma carbon bond.
>>> ie
>>> CH3
>>> /
>>> Ca-Cb-HCg
>>> \
>>> CH3
>>>
>>> instead of something more reasonable like...
>>>
>>> CH3
>>> /
>>> Ca-Cb-Cg-H
>>> \
>>> CH3
>>>
>>>
>>> Attached are the before and after pdb files. If these hydrogen
>>> placements are left in the current position the miniziation will fix
>>> them but often requires MAJOR side motions...
>>>
>>> I haven't seen this behavior in older standalone psfgen use.
>>>
>>> Brian
>>>
>>>
>>> ************************************************
>>> Brian Bennion, Ph.D.
>>> Biosciences Directorate
>>> Lawrence Livermore National Laboratory
>>> P.O. Box 808, L-448 bennion1_at_llnl.gov
>>> 7000 East Avenue phone: (925) 422-5722
>>> Livermore, CA 94550 fax: (925) 424-6605
>>> ************************************************
>>
>>
>
> ************************************************
> Brian Bennion, Ph.D.
> Biosciences Directorate
> Lawrence Livermore National Laboratory
> P.O. Box 808, L-448 bennion1_at_llnl.gov
> 7000 East Avenue phone: (925) 422-5722
> Livermore, CA 94550 fax: (925) 424-6605
> ************************************************