Dear Jason,
Thanks for the clear and quick answer - this makes sense now. Is there therefore a way of outputting the dihedrals combinations defined by the protein topology in a single line format in AMBER, such that each one only appears once? This is mainly for comparison reasons, as I would like to compare how the glycoprotein topology is defined in AMBER to how it is defined in a gromacs topology (both using the Amber ff99SB forcefield).
Many thanks,
Nathalie
________________________________________
From: Jason Swails [jason.swails.gmail.com]
Sent: 08 December 2015 15:55
To: AMBER Mailing List
Subject: Re: [AMBER] xparmed.py
On Tue, Dec 8, 2015 at 10:23 AM, Nathalie Willems <
nathalie.willems.bioch.ox.ac.uk> wrote:
> Dear Amber mailing list,
>
> I am using xparmed.py to output a list of all the dihedrals in a
> glycoprotein model I built using the GLYCAM glycoprotein builder tool. I
> start the xparmed program and execute these commands:
>
> xparmed.py
> choose the glycoprotein prmtop file (generated by tleap)
> click printDihedrals and choose all the atoms of the protein in the mask
>
> The list of dihedrals generated however, mention some strange atom names
> that are not originally present in the protein structure pdb file e.g 3C
>
> 16 C ( C) 18 N ( N) 20 CA ( CX) 22
> CB ( 3C) 2.0000 1.0000 0.0000 1.2000 2.0000
> M 16 C ( C) 18 N ( N) 20 CA ( CX) 22
> CB ( 3C) 1.8000 2.0000 0.0000 1.2000 2.0000
> M 16 C ( C) 18 N ( N) 20 CA ( CX) 22
> CB ( 3C) 0.8000 3.0000 0.0000 1.2000 2.0000
> M 16 C ( C) 18 N ( N) 20 CA ( CX) 22
> CB ( 3C) 0.0000 4.0000 0.0000 1.2000 2.0000
>
> It therefore also outputs dihedrals defined by the same 4 atoms. Looking
> back at the prmtop file for the protein, the atoms are defined like this:
> N H1 H2 H3 CA HA CB HB2 HB3 CG HG2 HG3 CD OE1 OE2 C O N H
> CA
> HA CB HB CG1 HG11HG12HG13CG2 HG21HG22HG23C O N H CA HA CB
> HB2 HB3
> OG HG C O N H CA HA CB HB2 HB3 CG HG2 HG3 CD OE1 NE2
> HE21HE22C
>
> Is it possible there is a bug in the xparmed.py program that results in
> splitting the columns of the prmtop file such that the program thinks there
> are atom types that don't really exists e.g 3C and 2C? Is this why there
> are multiple dihedrals generated for the same 4 atoms?
>
​While there are undoubtedly some bugs in ParmEd (as there are in any
non-trivial program), this is not one of them. The first character of each
line is either blank, M, or I. M originally stood for "multi-term", but
really means that 1-4 interactions for that pair are excluded (cpptraj
prints out "E" in this column to indicate End-group interactions are
omitted). That occurs *either* when a torsion is defined by multiple
Fourier terms (as is the case here), or in ring systems where 1-4 pairs are
connected by either bonds, angles, or other torsions (this has to do with
how the 1-2, 1-3, and 1-4 exclusions and exceptions are tracked by Amber,
and is beyond the scope of this email). An I indicates that this torsion
defines an "improper" torsion, which has a different bonding pattern in
general than proper torsions (typically there is one atom that is bonded to
the other 3).
Then, each of the 4 atoms involved in that torsion is listed in the format
number name (type name)
The type name is an implementation detail of the force field. The block of
the prmtop file you printed is the atom name (C, N, CA, and CB). The type
names (C, N, CX, and 3C) are defined in the AMBER_ATOM_TYPE section. Note
that there is no correspondence between the atom types and atom names in
general (although in practice the C, N, O, and CA atoms are very similar
for all amino acids, so there often appears *some* correspondence). The
importance of the type name is that all parameters between atoms that are
assigned the same types will be the same (unless you change one by hand
using ParmEd, for instance). This is the concept of "force field
transferability". Names are for all intents and purposes arbitrarily
defined within each residue (the naming convention is used from the PDB).
Types are names applied to classes of atoms that are expected to have
similar chemical characteristics. So if two atoms have the same atom type,
that means the force field thinks they are the same "kind" of atom (e.g.,
all carbon atoms that are sp3-hybridized bonded only to other C and H atoms
might be one type, sp2 carbons in a ketone group might be another, etc.).
The numerical columns after this are the force constant, periodicity, and
phase angle (followed by the 1-4 electrostatic and 1-4 van der Waals
scaling factors). So the block of text from xparmed.py that you printed
out indicates that there are 4 terms in the C-N-CA-CB torsion (that is
defined between atom types C, N, CX, and C3) -- those with periodicities 1,
2, 3, and 4 (the term with periodicity 4 has a force constant of 0, so it
does not contribute to the potential energy). So the potential energy
function of this torsion is a sum of 4 cosines (one of them zeroed), which
is why this torsion appears 4 times in the ParmEd output.
HTH,
Jason
--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Dec 08 2015 - 08:30:03 PST
