The keyword you need is FixQ (alias fixq). The input will look like this:
&fitq
(...)
FixQ :1.CA 0.20
FixQ :1.HA 0.10
FixQ :1.N -0.30
(...)
&end
Fixing the charges should also help to reduce the size of your matrix--if I
programmed it right, and I think I did, the fixed charges will not get
their own matrix columns but rather recede into the background. Another
option you may want to consider comes in the form of the "Tether" and
"TetherWeight" keywords. As implied, you can tether the resulting charges
to the initial topology, so if you have charges that you like for any
reason, you can supply those as an initial guess and the restraints will
pull the charges towards those values as opposed to zero.
When in doubt, always type "mdgx -FITQ" or "mdgx -YOUR_MODULE_NAME" on the
command line. It will print a formatted reference for all* keywords in
that module.
Also note that the size of the matrix will not just be the numbers of atoms
in all molecules--if you have many set to contain equal charges then this
will drastically cut down on the matrix size. Make sure you are setting
the maxmem / MaxMemory keyword to the actual amount of memory in your
system--it's got a fairly low setting by modern standards owing to this
module having come about back in 2011 or so. You can say "MaxMemory 16GB"
and things like that--no need to type a big ugly number and worry about the
number of zeros you included. Another thing to be aware of is the number
of fitting points that you are taking per molecule. The matrix will come
to (number of unique charges) x (number of data points in all molecules +
number of charge value restraints + number of charge sum restraints). The
numbers of restraints pale in comparison to the number of electrostatic
potential fitting points, but you can set the density, selection, and total
number of points with various keywords from the &fitq namelist. FitPoints
is the upper limit per molecule conformation, but ProbeSig, ProbeEps,
ProbeArm, and StericLimit are also pertinent to which points near the
molecular surface are selected.
Dave
* All keywords that I remembered to add to the glossary feature. I'm sure
I forgot one or two but there's always the reflector.
On Wed, Feb 24, 2021 at 8:16 AM David Cerutti <dscerutti.gmail.com> wrote:
> The keyword you need is FixQ (alias fixq). The input will look like this:
>
> &fitq
> (...)
> FixQ :1.CA 0.20
> FixQ :1.HA 0.10
> FixQ :1.N -0.30
> (...)
> &end
>
> Fixing the charges should also help to reduce the size of your matrix--if
> I programmed it right, and I think I did, the fixed charges will not get
> their own matrix columns but rather recede into the background. Another
> option you may want to consider comes in the form of the "Tether" and
> "TetherWeight" keywords. As implied, you can tether the resulting charges
> to the initial topology, so if you have charges that you like for any
> reason, you can supply those as an initial guess and the restraints will
> pull the charges towards those values as opposed to zero.
>
> When in doubt, always type "mdgx -FITQ" or "mdgx -YOUR_MODULE_NAME" on the
> command line. It will print a formatted reference for all* keywords in
> that module.
>
> Dave
>
> * All keywords that I remembered to add to the glossary feature. I'm sure
> I forgot one or two but there's always the reflector.
>
>
>
> On Wed, Feb 24, 2021 at 7:25 AM David A Case <david.case.rutgers.edu>
> wrote:
>
>> ----- Forwarded message from Luis Simon <lsimon.usal.es> -----
>>
>> Date: Wed, 24 Feb 2021 09:34:30 +0100
>> From: Luis Simon <lsimon.usal.es>
>> To: amber.ambermd.org
>> Subject: [AMBER] MDGX charge fitting with some charges fixed or restrained
>>
>> Dear Amber users:
>>
>>
>> I wonder if there is any way of running mdgx charge fitting specifying a
>> fixed charged for some of the atoms or group of atoms.
>>
>> “MinimizeQ" keyword request to fix all the charges restraining those of
>> the group of atoms specified by the following atom mask to zero… is ti
>> possible to restraint it to a different value than 0?
>>
>> I am trying to parametrize a set of molecules and I want to share the
>> same charges with some of the atoms (just like amino acid backbone atoms
>> share the same charges). MDGX allows this with the use of equalq keyword
>> and the use of atom masks, but since the charges for all the molecules must
>> be fitted simultaneously, my system run out of memory. A possibility would
>> be to fit only a subset of the molecules (or even different subsets and use
>> averages) and then transfer the charges of the backbone atoms to the rest
>> of the molecules, for which the charges will be fitted separately. Is it
>> possible to transfer them, fixing the charges of the common atoms?
>>
>> Thanks
>>
>> Luis
>>
>>
>>
>>
>>
>>
>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>>
>>
>> ----- End forwarded message -----
>>
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Feb 24 2021 - 05:30:03 PST
