AMBER Archive (2009)

Subject: Re: [AMBER] FF Parameters Transformations ...?

From: Karl Kirschner (kkirsch_at_scai.fraunhofer.de)
Date: Mon Apr 06 2009 - 02:39:31 CDT

Hi Marek,

        There is no easy solution for setting up your silicon compound. The best
solution would be to develop your own parameters, which then would give you a
feeling for where the weaknesses are in the resulting force field. This in
turn will allow you to judge if there are artifacts in the resulting
simulation that arise from the force field.

        Apart from that you can try to find existing parameters in published force
fields and try to transfer them to AMBER. We are doing this to some extent,
but have found that there are several small "pit falls" due to differences in
the force field equations and the various simulation parameters that each
program gives as default. If you do this, I suggest that you verify that the
parameters are transferred to AMBER correctly by running minimizations (check
each energy term outputted by each program) and simulations (check to see if
you capture the dynamics correctly) on several different traditional
molecules (hydrocarbons, alcohols, etc). Once you trust your parameter
conversion, then move onto your more exotic Si compound.

        I would be very cautious in using related carbon parameters, and then scaling
them. I am not sure a single or even multiple scaling factors for different
terms would capture the physics properly. This implies a linear relationship
in parameters for elements, and I have not personal seen that in my work.

        Just remember that the results of your simulation are directly connected to
the parameters that you use. If you use bad parameters, you will likely have
bad results.

Cheers,
Karl

On Wednesday 01 April 2009 19:29, Marek MalĂ½ wrote:
> Dear Amber users,
> some time ago I sent to this forum the
> below attached email regarding to calculations of Si-based molecules in
> Amber.
> Unfortunately I didn't receive any response till now :((
> Probably it is not so easy to answer/help here as it is clear to me from
> what
> I read about similar problems in this forum.
>
> So I have to this story just 2 additional questiones ( hopefully not so
> complicated as the original one :)) )
>
> #1 - as it is clear antechamber/parmchk routine has problem to associate
> Si atoms
> with the proper bond, angle, dihedral parameters since Amber
> forcefileds do not
> contain this parameters for silicon so it is necessary to look for it
> somewhere
> elsewhere and modify properly relevant FRCMOD file. I found this
> parameters in
> CVFF a PCFF forcefields so I will try to transport/transfer them for
> the Amber calculation. Is it a good idea ? What I should to care
> about ?
> Probably to check the physical units, check if mathematical
> expressions for
> the given type of forces is the same i both forcefields, if not make
> some relevant transformation
> of the transfered parameters ?
>
> Maybe could be a good/elegant idea to use well known
> parametrisations/parameters (like for example C-N, C-N-C, C-C-N-C)
> which are present in CVFF, PCFF but also in Amber ff to calculate
> some scaling constant/s which can be used for
> easy transformation of C-Si, C-Si-C ... parameters from CVFF, PCFF
> into Amber ff. Could be this a good idea ?
>
>
> Another possibility is to use PARMCAL routine - I have never used it
> till now so it would the case to use it to
> obtain at least the bond length and angle ff parameters for C-Si,
> Si-O, C-C-Si, C-Si-O ... bonds/angles ?
>
> Any experiences with this ffx->ffy transformations ... are highly
> welcomed ! I think that this issue could be a very interesting
> also for the other Amber users.
>
> #2 - On contrary it seems to me that antechamber/difcon ... has absolutely
> no problems with "exotic" Si atoms
> and for example calculated proper(?) Si partial charges. In PREPIN
> file (please see attached file), there is
> no error reported so I just would like to ensure: Is it from this
> point of view everything really OK - are the calculated
> Si charges usable ? Also PREPIN file contains info about bond
> lengths, angles, dihedral angles but this are stricly related
> to the actual values of the input molecular structure not the true
> equilibrium ones am I right ?
>
>
> Thanks a lot to anybody for any info or own experience with solving
> of similar problem or just for reference
> to relevant information resources (is there for example any
> apprpriate tutorial available ?) !
>
> Marek
>
>
>
> ####-AN---ORIGINAL---EMAIL
>
> ------- PÅ™edanĂ¡ zprĂ¡va -------
> Od: "Marek MalĂ½" <maly_at_sci.ujep.cz>
> Komu: amber_at_ambermd.org
> Kopie:
> Předmět: [AMBER] Parametrisation of Silicon based molecules for Amber
> simulations ?
> Datum: Fri, 20 Mar 2009 00:49:21 +0100
>
> Dear Amber users,
>
> I would like to simulate molecules which contain also Si (silicon) atoms.
> Please see attached fig. "mol01.jpg" or "mol.pdb."
>
> I tried to parametrise small "testing" molecule using Antechamber and
> I have obtained relevant PREPIN ("mol.prepin") and FRCMOD ("mol.frcmod")
> files.
> The prepin file is OK and also "Si" atoms were asigned with calculated
> partial charge etc.
>
> But there is problem with FRCMOD file which should be probably interpreted
> in this case only as
> an empty template for proper forcefield parameters (bond, angle, dihedrals
> parameters) which
> need to be included "manually".
>
> Nevertheless I tried to make PRMTOP and INPCRD file (just for the
> curiosity) and I succeeded,
> but after reading of this two files by "Chimera" and labeling the element
> types, the "Si" atoms
> were asigned with the "LP" label (please see "mol02.jpg") although in
> PRMTOP file are all "Si" atoms
> named properly and in section "FLAG AMBER_ATOM_TYPE" is clearly written
> type "Si". But this is probably
> question for "Chimera" forum :))
>
> It is clear to me that I will have to find a proper forcefield parameters
> so
> I am kindly asking for any info, links to relevant databases etc.
>
> For example I succeed to simulate this molecule in vacuum with MS software
> using "Dreiding" forcefield.
> Could be a good idea to adopt desiderative parameters from this forcefield
> and incorporate them into my FRCMOD
> file ?
>
> How to decide between for example two slightly different parametrisations
> which I could find in different data sources ( articles, databases ) ?
> Is there any highly recommended forcefield database available ?
>
>
> I appreciate any help and advices especially from the experienced
> researchers.
>
>
> Thank you very much in advance !
>
> Marek

_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber