Sage 2.1 t77-t79 cis amides

History

From Paul Labute at CCG:

The t77-t79 rules deal with amides:

[*:1]-[#7X3:2]-[#6X3$(*=[#8,#16,#7]):3]~[*:4] 0.2956 0 2.1697 180 0.0 0 0.0 0 # t75
[#1:1]-[#7X3:2]-[#6X3:3]=[#8,#16,#7:4] 1.2705 0 0.6343 180 0.0 0 0.0 0 # t76
[*:1]-[#7X3:2]-!@[#6X3:3](=[#8,#16,#7:4])-[#6,#1] -0.5339 0 2.5120 180 0.0 0 0.0 0 # t77
[#1:1]-[#7X3:2]-!@[#6X3:3](=[#8,#16,#7:4])-[#6,#1] 1.2104 0 1.3536 180 0.0 0 0.0 0 # t78
[*:1]-[#7X3:2]-!@[#6X3:3](=[#8,#16,#7:4])-[#7X3] 0.7926 0 0.9750 180 0.0 0 0.0 0 # t79

The t77-t79 specify chain bonds between the C=O and NX3, but if you look at the parameters they will encourage cis amides. The rules make a little more sense if you change the !@ (chain) to @ (ring) leaving t75 and t76 to deal with chains; t75-t76 encourage trans amides, as expected. Is possible that "!@" should be "@"?

Pavan:

Hyesu introduced the new amide terms in Parsley 1.3.0 (t77-79 were t70b,c,d in prior nomenclature) to deal with dialkyl amides, here is the release notes (https://openforcefield.atlassian.net/wiki/spaces/FF/pages/723025921 ).

Trevor:

We might want to forward this to Chapin or ask for his FF since he has a set of parameters for the biopolymer FF that deal with peptide torsions and it might have a different (better) prescription here. However, looking at this at face value, I'd say it is not necessarily encouraging cis because the extra [#6,#1] means it's only cis if we assume the match is #6, but flipped if the match is #1. If we assume a larger molecule and it is #6, then it does encourage cis and therefore flat/linear peptides; it is however the weakest torsion of all of them, and the periodicity 2 terms will dominate somewhat. I'd therefore agree with the broad response of "the values depend on the data we fit". It seems he is trying to take the k values and guess what molecules we fit to, or at least fix the smarts based on the k values. The ring parameters t75 and t76 do encourage trans because, well, they are in rings and limit certain geometries e.g. where the #1 can be (cis).