VMD-L Mailing List

From: Daniel Fellner (dfel694_at_aucklanduni.ac.nz)
Date: Thu Mar 21 2024 - 18:25:16 CDT

I'm having a similar issue right now. I included all the non-hydrogen
analogous dihedral terms for optimisation but two of them simply won't fit
the QM PES no matter what.

I think the problem is an ill-fitting angle/bond or two, so am repeating
the Hessian bond/angle optimisation with the latest dihedral parameters. If
that doesn't resolve it, I will switch to ffparam-gui and fit the
bonds/angles to their QM PES instead.

*Daniel Fellner BSc(Hons)*
PhD Candidate | Research Assistant
School of Chemical Sciences
University of Auckland
Ph +64211605326

On Fri, 22 Mar 2024 at 02:27, francesco quilli <
quilli.1855253_at_studenti.uniroma1.it> wrote:

> Thank you all for the prompt responses and the various approaches to the
> problem.
>
> As advised by you, I have researched the dihedrals related to the one to
> be optimized.I used the Torsion Explorer tool, which allows for trajectory
> analysis. From there, I selected several dihedrals associated with the
> first minimum of the PES that I am unable to fit. I performed the
> calculations and conducted several optimization attempts, but they didn't
> lead to significant improvements. Overall, the RMSE is slightly lower, but
> none of the MM profiles reach that minimum. I really appreciate your help
> and attention to my issue.
>
> Best regards,
>
> Francesco Quilli
>
> Il giorno ven 15 mar 2024 alle ore 00:23 Klimkowski, V <
> vklimkow_at_butler.edu> ha scritto:
>
>> It might be worth 1st getting it out of the spiro compound, refining
>> the dihedrals & then using these refined parameters as the starting point
>> to rerefine the spiro compound. Specifically breaking the CH2-CH2 bond in
>> the dioxolane, & refining the CH2-C-O-CH3 dihedral.
>>
>> Additionally, look at this article for some ideas:
>> https://urldefense.com/v3/__https://doi.org/10.1007/s10822-022-00464-3__;!!DZ3fjg!8AAYUf0y405AuuPArueCo6lwEyD1Q_1FGwUiNK1nVCezS8B953SWv3NxhLywzrQb5UKq7uq3lCUiV5ErTk1J5a2jUDQ$
>> See what special considerations they used to refine these fused ring
>> compounds.
>>
>> Interesting Problem. Best,
>> V Joe Klimkowski, PhD
>> Chemistry Research Affiliate
>> Butler University
>> Indianapolis, IN
>> <https://urldefense.com/v3/__https://doi.org/10.1007/s10822-022-00464-3__;!!DZ3fjg!8AAYUf0y405AuuPArueCo6lwEyD1Q_1FGwUiNK1nVCezS8B953SWv3NxhLywzrQb5UKq7uq3lCUiV5ErTk1J5a2jUDQ$ >
>> On the force field optimisation of $$\\beta$$ β -lactam cores using the
>> force field Toolkit - Journal of Computer-Aided Molecular Design
>> <https://urldefense.com/v3/__https://doi.org/10.1007/s10822-022-00464-3__;!!DZ3fjg!8AAYUf0y405AuuPArueCo6lwEyD1Q_1FGwUiNK1nVCezS8B953SWv3NxhLywzrQb5UKq7uq3lCUiV5ErTk1J5a2jUDQ$ >
>> When employing molecular dynamics (MD) simulations for computer-aided
>> drug design, the quality of the used force fields is highly important. Here
>> we present reparametrisations of the force fields for the core molecules
>> from 9 different $$\\beta$$ β -lactam classes, for which we utilized the
>> force field Toolkit and Gaussian calculations. We focus on the
>> parametrisation of the dihedral angles, with the goal of reproducing the
>> optimised quantum geometry in MD simulations. Parameters taken from CGenFF
>> turn out to be a good initial guess for the multiplicity of each dihedral
>> angle, but the key to a successful parametrisation is found to lie in the
>> phase shifts. Based on the optimised quantum geometry, we come up with a
>> strategy for predicting the phase shifts prior to the dihedral potential
>> fitting. This allows us to successfully parameterise 8 out of the 11
>> molecules studied here, while the remaining 3 molecules can also be
>> parameterised with small adjustments. Our work highlights the importance of
>> predicting th
>> doi.org
>>
>> ------------------------------
>> *From:* owner-vmd-l_at_ks.uiuc.edu <owner-vmd-l_at_ks.uiuc.edu> on behalf of
>> Gumbart, JC <gumbart_at_physics.gatech.edu>
>> *Sent:* Thursday, March 14, 2024 8:18 AM
>> *To:* Daniel Fellner <dfel694_at_aucklanduni.ac.nz>
>> *Cc:* francesco quilli <quilli.1855253_at_studenti.uniroma1.it>; Ernesto
>> Aleksei Delgado Hurtado <ERDELGADO_at_udec.cl>; VMD Mailing List <
>> vmd-l_at_ks.uiuc.edu>
>> *Subject:* Re: vmd-l: FFTK - Dihedral doesn't fit the QM target data
>>
>> Agreed, there may be be a missing parameter or two that’s leaving it
>> unable to fit that first minimum.
>>
>> Best,
>> JC
>>
>> On Mar 14, 2024, at 4:51 AM, Daniel Fellner <dfel694_at_aucklanduni.ac.nz>
>> wrote:
>>
>> There may be other correlated dihedrals which also require simultaneous
>> fitting, even if they have low penalty scores, this usually fixes the
>> problem for me.
>>
>> *Daniel Fellner BSc(Hons)*
>> PhD Candidate | Research Assistant
>> School of Chemical Sciences
>> University of Auckland
>> Ph +64211605326
>>
>>
>> On Thu, 14 Mar 2024 at 21:06, francesco quilli <
>> quilli.1855253_at_studenti.uniroma1.it> wrote:
>>
>> Thanks for the prompt responses,
>>
>> The dioxa azaspiro group is already a fragment of a larger molecule; the
>> dihedral of interest is located at the center of this group, and it is
>> impossible to fragment it further.
>>
>> I analyzed the rotation using the Torsion Explorer tool, and the
>> generated conformations do not seem to present steric clashes.
>>
>>
>> As recommended, I conducted tests with a smaller step size: Scan +-90°
>> Step size 5° and Scan +-90° Step size 3°. As can be observed from the PES
>> scans (PES profiles are shown in CG321_CG321_CG3C50_OG3C51.pdf), not much
>> has changed since the previous tests (Scan +-90° Step size 15° and Scan
>> +-90° Step size 10°), and the problem persists. I would appreciate your
>> advice on what I could try to improve these PES profiles.
>>
>>
>> Best regards,
>> Francesco
>>
>>
>> Il giorno mar 12 mar 2024 alle ore 03:43 Gumbart, JC <
>> gumbart_at_physics.gatech.edu> ha scritto:
>>
>> Indeed, it’s possible that steric clashes are interfering with the
>> rotation, which creates those somewhat unusual PMFs. But at the very
>> least, decreasing the step size is a good idea. The PMFs look rough.
>>
>> Best,
>> JC
>>
>> On Mar 11, 2024, at 6:21 PM, Ernesto Aleksei Delgado Hurtado <
>> ERDELGADO_at_UDEC.CL> wrote:
>>
>> Hi francesco
>> I can tell by the shape of your PES scans that you are trying to fit your
>> parameters using the whole molecule.
>> The recommended approach is by dividing your molecules into smaller
>> pieces that contain your functional groups of interest, fitting your
>> parameters on those molecules and then copying the parameters to use on
>> your whole molecule.
>> Regards,
>> Ernesto Delgado
>> El 11/3/24 a las 12:12, francesco quilli escribió:
>>
>> Hi all,
>>
>> I am currently working on resolving a dihedral belonging to a dioxa
>> azaspiro group. I have generated the QM target data, but I am encountering
>> difficulties in fitting the dihedral to the target.
>> I utilized the CGenFF program to identify entities with high penalties that
>> require further validation/optimization. Based on the output of CGenFF:
>>
>>
>> dihedral {CG321 CG321 CG3C50 OG3C51} K=3.4000 Periodicity 1 Phase shift=
>> 180° Penalty=53;
>>
>>
>> I initiated a ±90° scan with a 15° step size. I attempted various
>> combinations (i.e., periodicities, phase shift angles, and optimization
>> algorithms as you can see in CG321_CG321_CG3C50_OG3C51.pdf), but I am
>> unable to reduce the RMSE beyond 1.769. Additionally, in the target data, I
>> observe a first minimum that is not reached in any of these profiles (PES
>> profile are shown in CG321_CG321_CG3C50_OG3C51.pdf), despite trying
>> different combinations, as mentioned above, I also attempted a ±90° scan
>> with a lower step size (10°), but the results are very similar.
>>
>> I find myself stuck in this situation. Could anyone provide advice or tips
>> on this issue? Should I consider lowering the step size even further? I
>> would greatly appreciate any guidance.
>>
>> PS: I'm using VMD 1.9.4 and ORCA 5.0.4.
>>
>> Best,
>> Francesco
>>
>>
>>
>>
>>
>>