[gmx-users] Preparing a silicon bulk

Justin Lemkul jalemkul at vt.edu
Fri Dec 19 13:14:57 CET 2014

On 12/18/14 11:46 PM, Kevin C Chan wrote:
> Due to some reasons I could not receive digests from the list so I could
> only copy and paste from the archives and reply to you. I really hope this
> reply could go to the same thread of my question so it would not screw the
> list archive.

Nope, this is the start of a new thread now.  Make sure you sort out whatever 
delivery issues you're having using the admin page.

> Back to the question, first thank you so much for your reply Justin. I
> understand that pdb2gmx might not be useful, I just tried it. g_x2top seems
> better but I could not get it work even following the file format: it says
> it could find 1 name to type translations but it can not find forcefield
> for atoms like
> Opening force field file
> /home/kevin/opt/gmx_fftw3_double/share/gromacs/top/gromos54a7.ff/atomname2type.n2t
> There are 2 name to type translations in file gromos54a7.ff
> Generating bonds from distances...
> atom 8
> Can not find forcefield for atom SI1-1 with 0 bonds
> Can not find forcefield for atom SI2-2 with 0 bonds
> Can not find forcefield for atom SI3-3 with 0 bonds
> Can not find forcefield for atom SI4-4 with 0 bonds
> Can not find forcefield for atom SI5-5 with 0 bonds
> Can not find forcefield for atom SI6-6 with 0 bonds
> Can not find forcefield for atom SI7-7 with 0 bonds
> Can not find forcefield for atom SI8-8 with 0 bonds
> I gave the n2t like
> SI   SI         1     28.08   3    SI 2.352     SI 2.352      SI 2.352
>   ;2.352 is measured from my input coordinates

Gromacs uses nm for everything, not Angstrom.  Given that bonds are detected 
based on a 10% distance tolerance, this is the reason for the above messages.

> and just trying with 8 SI atoms and in the future I definite would like to
> have more.
> While waiting for your comments on this g_x2top problem, I would also like
> to have suggestions on manually writing the top file. As I don't think we
> need those tedious dihedrals for a silicon bulk, I though the topology
> should look rather simple:
> #include "gromos54a7.ff"
> ; I have looked into gromos54a7 and found parameters for SI so this include
> is enough am I right?
> [ moleculetype ]
> ; name SomeName
> SI 8
> [ atoms ]
> 1  SI  1  SI  SI1  1  charge1
> 2  SI  1  SI  SI1  1  charge2
> ...
> 8  SI  1  SI  SI1  1  charge8
> ; Maybe bonds and angles are also needed? I am not sure

You need to determine a suitable force field based on something that is 
published and validated to work.  I don't know what the source of Si parameters 
is for Gromos96, but people use different force fields in published work.  If 
you're wondering if bonds and angles are necessary (certainly the former, likely 
the latter, and maybe/maybe not on dihedrals, depending on the parametrization) 
then you need to back up and do some evaluation of parameters that are already 
available.  The output is only as good as the input, and Gromacs tools won't 
work (too much) magic to save you from mistakes.


> #include "spc.itp"
> [ system ]
> SomeName
> [ molecules ]
> SI 8
> SOL 100
> Is this the case?
> Thanks in advance,
> Kevin
> On 12/18/14 12:38 PM, Kevin C Chan wrote:
>> Dear Users,
>> I wish to simulate a Silicon bulk using Gromacs. I have created the bulk
>> using a plugin in VMD but I found difficulties when generating a top file
>> from this pdb file. I have tried pdb2gmx and x2top but both does not work.
>> Gromacs tells me that although we already have parameters for SI in maybe
>> ffnonbonded.itp, we do not have SI in the residue topology database and
> also
>> ffbonded.itp. How could it be and I do appreciate if anyone could share
> his
>> experience of simulating a silicon bulk in Gromacs.
> pdb2gmx is not the answer, since it was designed for linear polymers with
> limited branching.  The only option for materials like this is g_x2top.
> That
> requires creation of a suitable .n2t file, the format for which is
> described in
> http://www.gromacs.org/Documentation/File_Formats/.n2t_File - the contents
> should be fairly simple (one or two lines?) for silica.
> But...
>> Another thing, I would like to assign different charge to different SI
> atom
>> in my bulk. How can I achieve this?
> ...if you want to do this, there is no automated method.  You could, in
> theory,
> assign different residue names to each Si unit and assign that charge, but
> then
> the parameters of O are likely to be contingent upon the type of Si to
> which it
> is attached.  This eliminates g_x2top, which requires uniformity.  And as
> said
> above, this is not a task suited for pdb2gmx.  So either you need to (1) use
> g_x2top and hack or (2) write your own sort of code/script to generate the
> topology.
> -Justin


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