2021-03-26 WBO/Impropers meeting notes

Participants

@Jessica Maat (Deactivated)

@Trevor Gokey

@Christopher Bayly

@Simon Boothroyd

@Pavan Behara

Discussion

JM: Here are the molecules I made for impropers dataset based on our discussion in last meeting, with O-, Me, CN, NH3+ at the top of phenyl ring, and a pyridazine ring

SB/CB: A small correction, need to add the aniline group in para position

JM: Sure, will update that.

CB: Also, no ortho substituents like in the pyridazine ring, let’s not do the N-N in 4,5 positions yet, just stick with phenyl rings for simplicity to avoid strong sterics.

Another suggestion is to include 5-membered rings as Rowley did in his work.

SB: We are entering muddy waters if we include more, for the initial analysis may be we stick to some simpler series and expand that in step2 after looking at the results from this first set, working in small batches would be beneficial to identify bottlenecks, I do agree that there would be some deficiencies with this approach.

CB: Yeah, that’s a good first step, but we need to address some of the complexity in the first step itself, one way I could think of is considering NH2, NHMe, and NMe2 instead of just NH2 aniline, that way we would cover some of the complexities.

SB: Yeah, that’s a great idea to include primary, secondary and tertiary

CB: Minimally we will have, say 12 substituents covering the spectrum of electron donating to electron withdrawing on each row, and first row is for NH2, next for NHMe, and the last for NMe2.

In a similar fashion, we do for 5-membered rings too, a pyrrole with different substituents.

And with these datasets we can look at, improper for N, valence angles, and torsion parameters.

sp2_C - sp2_C - trivalent N - * (non-aromatic)

sp2_C - sp2_C - trivalent N - sp3_C (non-aromatic)

As far as substituents we can look at the Hammett table and pick some.

JM: Why NH2, NHMe, NMe2?

CB: This way we are keeping the electron donation and withdrawal effects from the phenyl group constant and varying the effect of proton and methyl groups on Nitrogen. Usually you don’t see many free anilines in drug-like compounds and knowing the consequences of having a methyl next door would add value to this work.

SB: JM and I discussed what to scan last week the valence angles along with the improper.

CB: Yeah, valence angle with N at center, may be 2 deg, 5 deg and 10 deg, and then dihedral angle with first substituent upto 90 deg. I remember JM was adding up the angles to see if the N is planar or not, so when the sum adds up to 360 then it is planar, and with 109.5 deg each it adds up to 330 deg (when pyramidal).

Another 3-dimensional thing would be, say we pick a linear scan of 1-D torsion at 45 deg for different substituents and pick up the valence angles from planar and perpetrate them onto the pyramidal and do a scan so that we would capture the deformations in energy landscape.

SB: That sounds pretty complicated, I think we would stick to 2D scan of dihedral+improper in the first step and then in step 2 do the perturbations to the pyramidalization of N.

CB: Okay, at least before choosing the substituents do a single geo opt and check where on the spectrum those different substituents lie, I know for sure that O- and other strong electron donating groups will be pyramidal and ammonium would be planar.

JM: I think some of the minima are accessible by both planar and pyramidal, so it’s tricky if we do that.

SB/CB: Yeah, that’s a good point.

TG: I have some questions about the periodicity of torsions, did you do any of this work before?

CB: I think I looked at the profiles manually and chose 1-fold or 2-fold based on that, in the end we actually end up with a dozen molecules or so which we analyze thoroughly before deciding a new parameter and the chemistry intuition helped me, may be you can try fourier analysis to extract the signals or akaike information criterion to check whether the splits, resulting in a new parameter being added, are helping or not in improving the objective function.

I remember once Chaya proposed a 6-fold term for torsions and it was not really necessary because the effect was due to sterics. If you consider a rotation around a sp3_C - sp3_C then it must be 3-fold, and when we have cis&trans we need a 1-fold+2-fold o/w we can’t differentiate between the two, and sometimes we get away with 1-4 vdW terms that describe the cis trans differences. So, looking at the chemistry helps.