Hi Will,
these things go back a long way so aren't really discussed much in recent
papers. I would take a look at the Cornell et al paper that underlies most
of the current variations (
https://pubs.acs.org/doi/pdf/10.1021/ja00124a002).
Look especially at "General features of the model".
For testing, MM bonds tend to be fairly rigid if you stick with the
harmonic model, so you could probably use the lengths and frequencies to
set the parameters (watch out for the lack of 1/2 in the Amber-version of
Hooke's law as well as in the angles and other terms, it is incorporated
into the "force constant". check the Cornell paper eq 1 for details).
For metals, people have gone beyond these simple descriptions by using
fancier functional forms, but I'm not an expert in that area so others here
may have some good advice.
carlos
On Mon, Mar 25, 2024 at 5:04 PM William Livernois <willll.uw.edu> wrote:
> Hi Carlos,
>
> Thanks for the prompt response, this is quite helpful! That makes sense
> given the context, do you know where on the website/manual this it
> outlined? I see the scaling of the 1-4 terms but I didn't see anything
> about bonded atoms not having 12-6 terms, do bonded atoms have a Coulomb
> energy applied the pair or is that also scaled/omitted?
>
> The context for this is that I am hoping to parameterize metal-ligand
> interactions using the Seminario method like in the MCPB.py toolbox. This I
> will be using QM calculations directly (projecting the Hessian Matrix onto
> the internal coordinates) to generate bonding parameters. So when you say
> that the QM data won't match the MM data I'm wondering what other checks I
> can do to make sure that the bonding parameters I generate are practically
> useful to generate MD trajectories. I was hoping that I could compare
> directly with a sample of bond distances to show the range where the
> second-order approximation holds, are there other approaches that would
> make more sense for this?
>
> Cheers,
>
> Will
>
> On Mon, Mar 25, 2024 at 1:38 PM Carlos Simmerling <
> carlos.simmerling.gmail.com> wrote:
>
>> Will, if you're only moving the H3, then the vdw changes should just
>> relate to pairs involving the H3. the MM force field doesn't include pairs
>> for which the H3 is involved in a bond or angle term with the other atom.
>> also, the "1-4"
>> ZjQcmQRYFpfptBannerStart
>> This Message Is From an Untrusted Sender
>> You have not previously corresponded with this sender.
>> See https://itconnect.uw.edu/email-tags for additional information.
>> Please contact the UW-IT Service Center, help.uw.edu 206.221.5000, for
>> assistance.
>>
>> ZjQcmQRYFpfptBannerEnd
>> Will, if you're only moving the H3, then the vdw changes should just
>> relate to pairs involving the H3.
>> the MM force field doesn't include pairs for which the H3 is involved in
>> a bond or angle term with the other atom.
>> also, the "1-4" terms (3 bonds away) are scaled.
>> There are probably better ways to get the energy data for those
>> structures, such as using cpptraj. I also would consider that even 1 step
>> of min will change the structure when it is so highly strained.
>> and overall... I'm not sure that bonds the resulting info will be
>> helpful. the harmonic MM bond won't really match up to real QM data in any
>> case.
>> are you trying to use the energies that you get, or just doing this to
>> make sure you understand the force field? My opinion would be that the
>> energies you get from bond stretching won't be useful. normally the
>> reference length and the force constant come from sources other than this
>> (like lengths from crystal structures, force constants from vibrational
>> frequencies).
>>
>> On Mon, Mar 25, 2024 at 4:25 PM William Livernois via AMBER <
>> amber.ambermd.org> wrote:
>>
>>> Hello,
>>>
>>> I am currently working on force field development for a modified DNA
>>> structure and I'm trying to get a handle on the force field
>>> implementation
>>> in AMBER. For testing purposes I have been sampling energies from the
>>> N3-H3
>>> bond in a terminated DTN residue. Following is my procedure:
>>>
>>> 1. I have generated the DTN Structure by placing the terminating OH
>>> hydrogen at (0,0,0) and importing the structure into tleap (all atoms
>>> are
>>> automatically added).
>>> 2. The BSC1 forcefield was applied to generate topology and coordinate
>>> files
>>> 3. From this structure the N3-H3 bond was stretched and contracted
>>> using
>>> PyTraj
>>> 4. I have used both the PyTraj/libsander interface and a simple 1-step
>>> minimization to confirm the energies at each step:
>>> 1-step energy
>>> &cntrl
>>> imin = 1,
>>> maxcyc = 1,
>>> ncyc = 1,
>>> ntb = 0,
>>> ntr = 0,
>>> cut = 9999
>>> /
>>> - *Note: this was also confirmed using the parm98 forcefield built
>>> into
>>> the Gaussian DFT package, which gave very similar values*
>>> 5. Finally, I plotted the energies from this and compared to the
>>> sum
>>> of the following energies based on manually calculating the N3-H3
>>> interaction by itself using BSC1 parameters:
>>> - Coulomb energies (using qN3 = -0.434, qH3 = 0.342)
>>> - Bonded energies (using k = 434 kcal mol^-1 Angstrom^-2 and r0 =
>>> 1.010)
>>> - Lennard Jones energies (using rmin = 0.5*(1.8240 +1.1870)
>>> Angstroms
>>> and eps = sqrt(0.1700*0.0157) kcal mol^-1 based on the mixing
>>> rules from
>>> the manual)
>>> - These parameters can be found on line 974 and 993 of parm10.dat
>>>
>>> I have found that the energies I get from AMBER match up well for the
>>> bonded energy curve, but the Lennard Jones curve I calculate causes a
>>> huge
>>> energy rise at lower N3-H3 bond lengths that doesn't show up in the
>>> calculations. This increase in energy from the LJ/VDW energy happens to
>>> match up with the trend of DFT energies at those same bond lengths. Am I
>>> missing a key parameter in AMBER that accounts for this difference or am
>>> I
>>> calculating the Lennard Jones parameters incorrectly? Or perhaps there is
>>> another better way to sample the energy from a force field?
>>>
>>> Regards,
>>>
>>> Will
>>>
>>> --
>>> William Livernois
>>> Dept. Electrical and Computer Engineering
>>> Email: willll.uw.edu
>>> Phone: 813-323-1920
>>> Pronouns: he/him/his
>>> _______________________________________________
>>> AMBER mailing list
>>> AMBER.ambermd.org
>>> http://lists.ambermd.org/mailman/listinfo/amber
>>> <https://urldefense.com/v3/__http://lists.ambermd.org/mailman/listinfo/amber__;!!K-Hz7m0Vt54!m-fo-qMQHdD-PvCaXRJEtbyFN0WBjwKIWbi0lRjCPiCI5HA9GaNoZ8oF-8nwWLfJuiyB9gE2VwsvriixGza8hlVf$>
>>>
>>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Mar 25 2024 - 15:30:04 PDT
