great work Dave!
carlos
On Wed, Jun 24, 2020 at 10:39 AM David Cerutti <dscerutti.gmail.com> wrote:
> Dear Developers,
>
> After some months in collaboration with Silicon Therapeutics, I am pleased
> to announce the first merge to master of code for supporting virtual site
> customization in Amber simulations. The pmemd and pmemd.cuda programs have
> been upgraded to detect two new fields in the Amber prmtop format,
> VIRTUAL_SITE_FRAMES and VIRTUAL_SITE_FRAME_DETAILS, that specify up to
> seven unique virtual site frames. (The former provides integer atom indices
> for the virtual site and its frame, the latter gives dimensions on the
> frame.) All of the frames described in the mdgx chapters of the Amber20
> manual are now supported by our main simulations engine, as well as two new
> additions: a two-site frame that places the virtual site at a fixed
> distance from its parent atom, and a four-site frame that can model objects
> like ammonia nitrogen lone pairs. All of the original virtual site
> support, inferring the types from bonding patterns, remains active in the
> program. Silicon Therapeutics is interested in using the virtual sites for
> their free energy calculations, so the new additions in both programs
> support TI through the "dual-topology" method as well.
>
> While the pmemd program now detects the new frames, making force fields
> with them is another matter. The ideal solution may be to have antechamber
> take directives to include virtual sites in the models it makes, and add
> information to the prepin / OFF lib formats that tLEaP can read. As CanDo
> becomes our authoritative topology and system creator, this functionality
> can go there as well. In the near term, however, the mdgx program can take
> a simple namelist-format file and apply the information to all residues in
> a standard Amber prmtop to create a new prmtop with VIRTUAL_SITE_FRAMES and
> VIRTUAL_SITE_FRAME_DETAILS fields. Here is an example of rules for turning
> all TIP3P waters in a simulation into TIP5P-Ew:
>
> % These rules add extra points to all waters in a prmtop
> &rule
> epname EP1
> eptype Out-3
> parent O
> frame2 H1
> frame3 H2
> residue WAT
> v12 -0.344908263
> v13 -0.344908263
> v12x13 0.644379035
> q -0.2410
> &end
>
> &rule
> epname EP2
> eptype Out-3
> parent O
> frame2 H1
> frame3 H2
> residue WAT
> v12 -0.344908263
> v13 -0.344908263
> v12x13 -0.644379035
> q -0.2410
> &end
>
> % These rules modify the existing TIP3P atoms to become TIP5P-Ew
> &rule
> atom O
> eptype NONE
> residue WAT
> q 0.0
> sig 3.09700
> eps 0.17800
> &end
>
> &rule
> atom H1
> eptype NONE
> residue WAT
> q 0.2410
> &end
>
> &rule
> atom H2
> eptype NONE
> residue WAT
> q 0.2410
> &end
>
> Using these &rule namelists, one can add virtual sites to ligands, protein
> residues, and any other system component. The mdgx program provides
> detailed documentation of how to use the &rule namelist, in addition to the
> new &parmedit namelist for rebuilding topologies. Type the name of any
> namelist (in all caps) as an argument to mdgx on the command line:
>
> mdgx -RULE
> mdgx -PARMEDIT
>
> A number of examples can be found in the mdgx test cases. While the
> topology writing feature of mdgx will be superceded by CanDo in time, its
> facility for fitting parameters for the new virtual site frames will remain
> Amber's primary means for developing these models. For years, mdgx has had
> the ability to add these customized virtual sites to a topology and fit
> partial charges, and I have done a number of calculations examining what
> they can do for ESP fitting and hydration free energies. More recently I
> have added support to the valence bond parameter module, but this has not
> yet seen a s application. In summary, all the tools are now in place to
> create Amber force fields with a diverse selection of massless sites. We
> can look forward to a leap forward in model development in the coming
> years.
>
> Dave
> _______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber-developers
>
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Wed Jun 24 2020 - 08:00:05 PDT
