Give mdgx a try for this:
http://ambermd.org/tutorials/advanced/tutorial32/index.php
That's the program that we are actively maintaining and upgrading with
time. It's been used to create an entire protein force field (330
peptides, 250,000 total structures), and an equally large extension for
non-natural amino acids. It can do individual drug molecules, anything
that you can make Amber topologies for. It also has a nice facility for
manipulating molecules that will help you create the data set:
http://ambermd.org/tutorials/basic/tutorial6/index.php
Controls are given in a format that mimics the sander &namelist syntax.
You can see the parameter fitting options for the ¶m namelist by typing
on the command line (assuming that $AMBERHOME is set and in your path):
mdgx -PARAM
As of Amber20, I've been trying to improve the output options, but it
sounds like you just want frcmod files which mdgx has been printing for
some time.
Dave
On Fri, Aug 14, 2020 at 6:12 PM Robin Betz <robin.robinbetz.com> wrote:
> Hi Samuel,
>
> You're correct that there shouldn't be a difference in how you use Paramfit
> with more or fewer structures.
> The problems you're seeing are coming from the operating system and your
> script.
>
> The "Killed" message is the out-of-memory (OOM) killer killing Paramfit as
> there isn't enough memory available. If you're running on a compute
> cluster, you can increase the memory available to the job. If you're
> running on your own machine, you can adjust the nice-ness of the process,
> use swap, killing other memory-intensive processes, etc.
>
> The "syntax error" message seems to be coming from bash, so I'm guessing
> you're using a script to call paramfit and somehow it's trying to execute
> the output of Paramfit, which gives syntax errors. If you attach your run
> script I can tell you more specifically.
>
> Hope this helps,
> Robin
>
> On Fri, Aug 14, 2020 at 2:58 PM Samuel Himes <sam.himes92.gmail.com>
> wrote:
>
> > Hello Amber Users,
> >
> > I am attempting to use Paramfit to generate parameters for molecules of
> > interest. In my initial trial I was able to run paramfit from start to
> > finish with 20 structures. I wanted to repeat the process but this time
> > with about 350 structures. This time I can only complete the first step
> to
> > fit K with that many structures. After that I use the following job
> control
> > file.
> >
> > # Run a fit to energies using the parameters defined earlier
> > RUNTYPE=FIT
> > COORDINATE_FORMAT=TRAJECTORY
> > NSTRUCTURES=364
> > K=-292501548281.682251
> > PARAMETERS_TO_FIT=LOAD
> > PARAMETER_FILE_NAME=prms.in
> > FUNC_TO_FIT=SUM_SQUARES_AMBER_STANDARD
> > QM_ENERGY_UNITS=HARTREE
> >
> > # Use the genetic algorithm with the following settings
> > ALGORITHM=GENETIC
> > OPTIMIZATIONS=50
> > MAX_GENERATIONS=10000
> > GENERATIONS_TO_CONV=20
> > GENERATIONS_TO_SIMPLEX=5
> > GENERATIONS_WITHOUT_SIMPLEX=5
> > MUTATION_RATE=0.100000
> > PARENT_PERCENT=0.250000
> > SEARCH_SPACE=-1.000000
> > SORT_MDCRDS=OFF
> >
> > # Output files for later analysis
> > WRITE_ENERGY=fit_output_energy.dat
> >
> > Every time I run it says “Killed” after about 30-40 minutes. Additionally
> > it throws these errors.
> >
> > -bash-4.2$ -350.1297 kcal/mol, Np = 2.0000, Phase = 180.0001 Deg
> > bash: -350.1297: command not found...
> > -bash-4.2$ (n -c3-nh-ce)*Kp = 351.1492 kcal/mol, Np = 2.0000,
> > Phase = 0.0000 Deg
> > -bash: syntax error near unexpected token `*Kp'
> > -bash-4.2$ (n -c3-nh-c3)*Kp = -23.1425 kcal/mol, Np = 2.0000,
> > Phase = 0.0000 Deg
> > -bash: syntax error near unexpected token `*Kp'
> > -bash-4.2$ (n -c -c3-nh)*Kp = 93.5440 kcal/mol, Np = 2.0000,
> > Phase = 180.0001 Deg
> > -bash: syntax error near unexpected token `*Kp'
> > -bash-4.2$ (n -c -c3-c3)*Kp = -73.0671 kcal/mol, Np = 2.0000,
> > Phase = 180.0001 Deg
> > -bash: syntax error near unexpected token `*Kp'
> > -bash-4.2$ (n -c -c3-c3)*Kp = 23.1595 kcal/mol, Np = 4.0000,
> > Phase = 180.0001 Deg
> > -bash: syntax error near unexpected token `*Kp'
> > -bash-4.2$ (nh-ce-ce-c2)*Kp = -219.8412 kcal/mol, Np = 2.0000,
> > Phase = 180.0001 Deg
> > -bash: syntax error near unexpected token `*Kp'
> > -bash-4.2$ (nh-c3-n -c3)*Kp = 146.5671 kcal/mol, Np = 2.0000,
> > Phase = 0.0000 Deg
> > -bash: syntax error near unexpected token `*Kp'
> > -bash-4.2$ (nh-c3-n -c )*Kp = 60.3180 kcal/mol, Np = 2.0000,
> > Phase = 0.0000 Deg
> > -bash: syntax error near unexpected token `*Kp'
> > -bash-4.2$ (nh-c3-c -o )*Kp = -14.6946 kcal/mol, Np = 2.0000,
> > Phase = 180.0001 Deg
> > -bash: syntax error near unexpected token `*Kp'
> > -bash-4.2$ (nh-c3-c3-ca)*Kp = 896.3533 kcal/mol, Np = 3.0000,
> > Phase = 0.0000 Deg
> > -bash: syntax error near unexpected token `*Kp'
> > -bash-4.2$ (c3-c3-nh-ce)*Kp = 843.2682 kcal/mol, Np = 2.0000,
> > Phase = 0.0000 Deg
> > -bash: syntax error near unexpected token `*Kp'
> > -bash-4.2$ (c3-c3-n -c3)*Kp = -24.0452 kcal/mol, Np = 2.0000,
> > Phase = 0.0000 Deg
> > -bash: syntax error near unexpected token `*Kp'
> > -bash-4.2$ (c3-c3-n -c )*Kp = 77.9657 kcal/mol, Np = 1.0000,
> > Phase = 180.0001 Deg
> > -bash: syntax error near unexpected token `*Kp'
> > -bash-4.2$ (c3-c3-n -c )*Kp = 28.4979 kcal/mol, Np = 3.0000,
> > Phase = 0.0000 Deg
> > -bash: syntax error near unexpected token `*Kp'
> > -bash-4.2$ (c3-c3-n -c )*Kp = 32.4247 kcal/mol, Np = 4.
> > -bash: syntax error near unexpected token `*Kp’
> >
> > The parameters that I’m setting to fit about 64 dihedral angles using the
> > KP setting.
> > Has anyone encountered a problem similar to this? Is it that I am simply
> > using too many structures, or is there a different problem? If the
> problem
> > has to do with the amount of structures how many structures are usually
> > necessary to create good parameters?
> >
> > Thanks in advance for the help.
> > -Sam
> > _______________________________________________
> > 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
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Aug 14 2020 - 16:00:02 PDT
