VMD-L Mailing List

From: Chitrak Gupta (chgupta_at_mix.wvu.edu)
Date: Thu Sep 03 2015 - 19:56:00 CDT

Hi Christopher,

Sorry, I ended up replying to you by mistake.

So, when I am dealing with an atom of the second category, what do I need
to do to parameterize the bonds, angles etc?

Regards,
Chitrak.

On Thu, Sep 3, 2015 at 12:20 PM, Mayne, Christopher G <cmayne2_at_illinois.edu>
wrote:

> Chitrak,
>
> Make sure that you continue to cc VMD-L so that others can learn from our
> discussion.
>
> ffTK does not support the NONBONDED (LJ) parameters from first principles;
> however, the values for these parameters MUST be provided in order to
> perform several of the optimization calculations. There are two cases:
>
> 1) the atom is close enough to a known atom type in the CHARMM/CGenFF
> parameter set.
> --Simply assign the atom type and include the CGenFF parameter files as
> the "associated parameter file."
>
> 2) the atom is not sufficiently close by analogy to any known types and
> must be assigned a unique atom type.
> --Since ffTK does not support the optimization of LJ parameters, you must
> assign the values of these parameters to the newly created atom type.
> Typically, you would find the closest match by analogy, and copy the
> parameters over. The BuildPar -> Assign Missing VDW/LJ Parameters by
> Analogy tool facilitates this process. The subtlety here is that you are
> selecting the LJ parameters by analogy, but will not match against any
> other parameters (bonds, angles, dihedrals) which you want to parameterize
> explicitly.
>
> Regards,
> Christopher Mayne
>
> On Sep 2, 2015, at 4:55 PM, Chitrak Gupta wrote:
>
> Hi Christopher,
>
> What I meant is this:
>
> The tutorial asks you to assign missing LJ parameters before the Gaussian
> optimization. My question is, do I HAVE to find a similar atom for each
> atom in my system? Or is it ok to leave a few of those blank if I can't
> find one?
>
>
> Regards,
> Chitrak.
>
> On Wed, Sep 2, 2015 at 1:29 PM, Mayne, Christopher G <cmayne2_at_illinois.edu
> > wrote:
>
>> Chitrak,
>>
>> Several ffTK features have evolved since the screencasts were made, so
>> some of the precise details between the screencasts and the current version
>> of ffTK have changed; however, the screencasts are still a good guide for
>> beginners.
>>
>> I'm a little unclear regarding your second question, but I'll take a
>> shot...
>>
>> I"m not quite sure I understand
>>
>> does all the non-bonded term have to be filled up before doing the
>> Gaussian Optimization?
>>
>> The Gaussian geometry optimization is only dependent on the PDB file.
>> Non-bonded parameters are only present in the parameter file. If you're
>> referring to atom types (which are related to non-bonded parameters), those
>> are in the PSF file, not the PDB file.
>>
>> For these atoms, should I look for the most similar ones, or should I
>> leave the non-bonded part blank?
>>
>> ffTK doesn't support parameterization of novel LJ parameters. You should
>> choose the closest you can find. There is a tool to help do something like
>> this: BuildPar -> Assign Missing VDW/LJ Parameters by Analogy, which is
>> what I think you are looking for.
>>
>> Regards,
>> Christopher Mayne
>>
>> On Sep 2, 2015, at 2:59 PM, Chitrak Gupta wrote:
>>
>> Hi Christopher,
>>
>> Thank you so much for your help. Yes, I am using VMD 1.9.2, and I had
>> missed the "Analyze" button (does the tutorial have it?). Once I went
>> through that step, everything worked out fine. I did add the all36_cgenff
>> and all36_prot PAR files in the "Associated parameter files".
>>
>> For my structure, I generated the pdb using Avogadro, loaded it into
>> Molefacture, and from that point just followed the tutorial. One question I
>> have is.......does all the non-bonded term have to be filled up before
>> doing the Gaussian Optimization? I am asking because one or two atoms in my
>> system are not exactly similar to the ones in the parameter files. For
>> these atoms, should I look for the most similar ones, or should I leave the
>> non-bonded part blank?
>>
>> Also, how would I include impropers like you suggested?
>>
>>
>> Regards,
>> Chitrak.
>>
>> On Wed, Sep 2, 2015 at 2:37 PM, Mayne, Christopher G <
>> cmayne2_at_illinois.edu> wrote:
>>
>>> Chitrak,
>>>
>>> I need a little bit more information in order to attempt to answer your
>>> questions.
>>>
>>> 1) Are you using VMD 1.9.2?
>>> 2) Are you including any "Associated Parameter Files"?
>>>
>>> In 1.9.2 we updated the process of building the initial PAR file, which
>>> includes a step to analyze the input molecule, cross-checking against the
>>> associated parameter files, and to visually inspect the missing
>>> parameters. When you click the "Analyze" button, the boxes just below
>>> should populate with only the missing parameters, which can be written out
>>> to file via the "Write Initial Parameter File" button just below that.
>>> Upon analysis, do any of the boxes populate with missing parameters?
>>>
>>>
>>> With regard to unnatural amino acids...
>>> I would treat the ends in the same manner that you anticipate using the
>>> amino acid in production simulations. i.e., if you are going to use the
>>> amino acid as a single small molecule, I would use charged ends; if the
>>> amino acid will appear in the middle of a peptide/protein, then I would
>>> apply standard types of patches for termini. I very much doubt that a
>>> protonated carboxylic acid is what you want. Furthermore, I would attempt
>>> to leave as much of the original amino acid parameters in tact, i.e., and
>>> only refine the portions that you have modified. I don't know how you have
>>> constructed your PSF/PDB input, but make sure that you include any required
>>> impropers.
>>>
>>> Regards,
>>> Christopher Mayne
>>>
>>>
>>> On Sep 2, 2015, at 11:05 AM, Chitrak Gupta wrote:
>>>
>>> > Hi!
>>> >
>>> > I am new to the FFTK plugin. I have followed the steps of the tutorial
>>> video to generate the initial psf file (and that has all the information).
>>> However, when I use it to generate the initial par file, none of the
>>> bond/angles etc are defined. It is essentially an empty file. I am
>>> copy-pasting the part of the par file below.
>>> >
>>> > I also have an additional question. I am trying to parameterize an
>>> unnatural amino acid. To make the molecule "complete", I have a -COOH as
>>> the backbone carbonyl, I have two HN atoms attached to the backbone N, etc.
>>> Would these make the parameterization inaccurate?
>>> >
>>> >
>>> > Here is my initial par file
>>> >
>>> > BONDS
>>> > !V(bond) = Kb(b - b0)**2
>>> > !
>>> > !Kb: kcal/mole/A**2
>>> > !b0: A
>>> > !
>>> > !atom type Kb b0
>>> > !
>>> >
>>> > ANGLES
>>> > !
>>> > !V(angle) = Ktheta(Theta - Theta0)**2
>>> > !
>>> > !V(Urey-Bradley) = Kub(S - S0)**2
>>> > !
>>> > !Ktheta: kcal/mole/rad**2
>>> > !Theta0: degrees
>>> > !Kub: kcal/mole/A**2 (Urey-Bradley)
>>> > !S0: A
>>> > !
>>> > !atom types Ktheta Theta0 Kub S0
>>> > !
>>> > !
>>> >
>>> > DIHEDRALS
>>> > !
>>> > !V(dihedral) = Kchi(1 + cos(n(chi) - delta))
>>> > !
>>> > !Kchi: kcal/mole
>>> > !n: multiplicity
>>> > !delta: degrees
>>> > !
>>> > !atom types Kchi n delta
>>> > !
>>> >
>>> > IMPROPER
>>> > !
>>> > !V(improper) = Kpsi(psi - psi0)**2
>>> > !
>>> > !Kpsi: kcal/mole/rad**2
>>> > !psi0: degrees
>>> > !note that the second column of numbers (0) is ignored
>>> > !
>>> > !atom types Kpsi psi0
>>> > !
>>> >
>>> > NONBONDED nbxmod 5 atom cdiel shift vatom vdistance vswitch -
>>> > cutnb 14.0 ctofnb 12.0 ctonnb 10.0 eps 1.0 e14fac 1.0 wmin 1.5
>>> > !
>>> > !V(Lennard-Jones) = Eps,i,j[(Rmin,i,j/ri,j)**12 - 2(Rmin,i,j/ri,j)**6]
>>> > !
>>> > !epsilon: kcal/mole, Eps,i,j = sqrt(eps,i * eps,j)
>>> > !Rmin/2: A, Rmin,i,j = Rmin/2,i + Rmin/2,j
>>> > !
>>> > !atom ignored epsilon Rmin/2 ignored eps,1-4
>>> Rmin/2,1-4
>>> > !
>>> >
>>> > END
>>> >
>>> >
>>> >
>>> >
>>> > Regards,
>>> > Chitrak.
>>>
>>>
>>
>>
>
>