2020-02-28 Chemical Perception meeting notes

Plots and data that were discussed during the call:

@Hyesu Jang's sixfold periodicity plots: [link]

@Chaya Stern (Deactivated) ’s presented WBO/pyramidal nitrogen plots: link to slides

Notes from video calls:

• Sixfold Periodicity discussion (HJ):

  • LPW: Only after adding six fold to periodicity of the torsion, the total energy (black line in Hyesu’s slides) will change shape. Six fold is the only way to improve accuracy of FF.

  • CB: Does toluene need six fold term?

    • LPW: Might need to give more attention to methyl rotations. Methyl rotations considered less pharmaceutically relevant. Try to make a case for chemically symmetric molecules, such as toluene.

  • DM: what’s driving the current interest in adding six fold term?

    • LPW: The tetrazole case. Hyesu could get the molecule to be planar but the PES shape is wrong. 

  • CB: How have we gotten away without adding this term for many years? Possibly bad practice historically in FF development (?)… CB wants to see an actual case where this works.

  • CS: Butane c-c-c-c term has a periodicity equal to 6. 6 fold has been done before. 

  • CS: What happens if we change the phase angle? 180, 0 etc.

    • LPW: There isn’t an effect (Displayed this on Hyesu’s spreadsheet)

  • CS: When doing Bayesian sampling to determine the periodicity, there were many ways to get the correct answer.

  • LPW: Would like to determine the answer in data-driven way. In the idealized case (shown Hyesu’s spreadsheet) the three fold and two fold cases perfectly cancel each other out. Surprising result. Maybe in a non-idealized case the 6 fold case might not be necessary.

  •  DM: Conclusion, we might need these somewhere. I am more interested in including these in the FFs.

  • CB: Agrees with DM. We might need the six fold terms. In a way, secretly hopes that we don’t need them because of additional complexity and added work. 

  • HJ: We might need six fold specifically for sp3 carbon and trivalent nitrogen torsion. May need to prove six fold term for the toluene case, will work on this future.

  • CB: Perfectly planar and perfectly pyramidal cases… what happens in the case of an intermediate case? 

    • HJ: Maybe we can use 6 fold and 3 fold torsion term to capture intermediate geometries. 

    • CB: Electron donating will affect the intermediate geometries. Maybe we need parameters that can scale the force constant to capture the intermediate geometries. 

  • DM: Let’s try testing the force field on benchmark set rather than compare objective function.

  • LPW: Agrees. Objective functional always increases with increased targets. 

• WBOs and pyramidal nitrogens (CS):

  • CS [shows slides with plots]: 

    • In general torsion barrier heights increase with ELF10 WBOs increase.

    • Bulky groups in torsion drives, there were odd peaks. 

    • The changes in the scans were coming from non-bonded interactions, and not the orbital effect that the torsion parameters are trying to capture. 

    • Looked at correlation between torsion scans the improper angle: There are flipping of nitrogen centers for molecules with lower WBO, less conjugation and more electron density over the nitrogen center.

  • CB: Is the bulky group always cationic?

    • CS: In most cases no. But there are sometimes oxygens. 

  • CB: Agree with non-bonded terms observation. The urea when rotating would have strong electrostatic effect. 

  • CS: I will look at subtracting out and to determine the contributions to the energy profile. Will look at residuals after subtracting out. Determine the non-bonded terms.

  • CB: This is a great job of detailing these effects in CS’s plots. Shows the potentials that we try to mimic with FFs. 

  • CS (going over plots):

    • Amino electron donating group on phenyl ring, the system does not conjugate. Lose symmetry in the scan

    • Molecule with urea on the meta position has improper angle flip

    • All molecules that have nitrogen involved in the torsion a flip have shown a torsion barrier flip if there is no conjugation

    • If there’s a conjugation then the molecule loses symmetry in the torsion drive profile, the improper angle shifts slightly but does not invert as intensely 

  • CB: This phenomena will be important for LPW and HJ torsion drives

  • LPW: We have been conscious of the fact that for the TD when we chose the lowest energy structure there are discontinuities in the orthogonal degrees of freedom. We knew this would happen but not sure what the plan is for taking care of it. If we did not have the option to parameterize the DOF that are causing this discontinuity. When relaxing in the MM, we might need to parameterize the improper nitrogens in the torsions.

Action Items: