Dear Prof. Simmerling,
Thank you for your response and indispensible info you provided.
IDIVF PK/6 PHASE PN
> > O-CT-N-C 1 0.382890 108.00 -1.
> >
>
> >it's not clear to me why you're dividing by 6.
>
> >(such as forgetting the factor of 2 in the amplitude, or getting the units
> wrong,
Well, I have thought PK would equal to half the potential energy at the
energy barrier [PK=(V2)/2]. That is
indeed what I did. PK/6 indeed means (V2)/2/6 as I have stated that in my
previous message as follows
"Instead, I took half the max potential-energies at the barriers and
divided them by 6".
I guess, I should have formulated my statement before for clarification ,
PK=(V2)/2.
Overall, I actually divided the potential energy at the barrier by 12 (V2 /
12).
Sorry for keeping you in dark for not providing the real equilibration I
used.
Is that what you were wondering? I would appreciate it much if you could
confirm the equilibration above?
> a negative amplitude here just inverts that term in the Fourier series,
>with a minimum instead of maximum.
That sounds interesting. I am currently doing a production run simulation
with the current parameters I found.
But, I also would like to try inverting the Fourier series for a minimum.
I am kind of in dark here though as I have not tried that before.
What do you mean exactly by "with a minimum"? Would that be the minimum
points right around the maximum point at the barrier ?
How would one invert the Fourier series?
>you might look at some of the papers, such as for the mdgx program, or
paramfit, or our paper
>on ff14sb (ff19sb is more complicated since it's 2D parameters).
I have looked at the ff19sb paper of yours. Well, that looks quite
advanced. Optimization of parameters
using a GRID model is quite interesting. It would be nice to learn that
algorithm for force field development.
How did you actually generate the GRIDs , using cpptraj or GAMESS-US?
Regards,
Cenk Andac
On Tue, Feb 8, 2022 at 12:55 AM Carlos Simmerling <
carlos.simmerling.gmail.com> wrote:
> I'll try to give you my thoughts on this, though there are many methods in
> use.
>
> IDIVF PK/6 PHASE PN
> > O-CT-N-C 1 0.382890 108.00 -1.
> >
>
> it's not clear to me why you're dividing by 6.
>
> (such as forgetting the factor of 2 in the amplitude, or getting the units
> wrong,
>
> > 1) It has been more than 10 years since I last parameterized a torsion
> > angle right. To refresh my memory, Can an amber expert possibly confirm
> the
> > general steps (outlined above) I took to derive the dihedral parameters.
> >
>
> it seems reasonable, but there are many ways to do this. You might look at
> some of the papers, such as for the mdgx program, or paramfit, or our paper
> on ff14sb (ff19sb is more complicated since it's 2D parameters).
>
> >
> > 2) In line of PN = -3, PHASE = -250.41, a negative PHASE value, does not
> > sound right to me? Can it be replaced with (360 - 250.41 =) 109.59 ?
> >
>
> the phase is just a shift in the position of the minimum and it shouldn't
> matter if you add 360 deg.
>
>
> > 3) I have checked some amber FF parameters and I noticed some negative PK
> > values. What is the rationale for negative PK values as energy barriers
> > possess positive energies in difference PES curves after zeroing the
> lowest
> > minimum energy point?
> >
> > a negative amplitude here just inverts that term in the Fourier series,
> with a minimum instead of maximum. This could be done to reduce a barrier
> that is too large. The same can be achieved with a different phase, so the
> details will depend on how the fitting was done. The rationale can vary
> depending on the protocol, but zeroing the lowest point isn't the only way
> to approach the fitting, so it doesn't always work as you outlined above.
>
> I would suggest redoing your grid scan to get energies using your new
> parameters, and then making sure that the resulting energy profile is an
> accurate match to your target profile. There are many ways it can go wrong
> (such as forgetting the factor of 2 in the amplitude, or getting the units
> wrong, or getting the divider wrong when you have multiple terms
> contributing to the bond rotational energy, or other details of the
> dihedral equation).
>
> Thanks in advance for enlightening me.
> >
> > Best,
> >
> > Cenk Andac
> > _______________________________________________
> > 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 Mon Feb 14 2022 - 01:30:03 PST
