Closed bannanc closed 7 years ago
Can you see whether GAFF/GAFF2 cover it, @bannanc ?
On the one hand, it seems like we should cover it, because "why not?" but on the other, I'm a little worried because I don't like either of the two most logical routes forward here: a) Assign a specific CO2 bond b) Modify the generic C=O so it uses [#6] rather than [#6X3] The former seems like being too specific, whereas I worry the latter might break something else.
Can you make C=C=O bonds?
On a similar issue there is also a C#O in the larger drugbank set (carbon monoxide)
The parameter usage takes 15 minutes to run, I'm waiting to get the gaff2 types to make it easier to look up what it would assign.
This doesn't seem to exist in GAFF2 though, so we should probably skip it.
GAFF2 for carbon dioxide:
c1-o 794.98 1.172 SOURCE2_SOURCE5 31 0.0068
c1 is an sp carbon
GAFF2 for the other C=O bonds:
c -o 652.57 1.218 SOURCE1_SOURCE5 27083 0.0110
c is specifically carbons in carbonyl groups
This is the only carbonyl group we have right now:
[#6X3:1]=[#8X1+0,#8X2+1:2] 570.0 1.229 parm99 carbonyl oxygen
This seems like a very different force constant anyway, so I'm not sure we should have anything specific here
@bannanc - I think skipping carbon monoxide for now makes sense: No need to introduce parameter complications for just one molecule (not covered by GAFF2 anyway).
In terms of carbon dioxide, you wrote:
This seems like a very different force constant anyway, so I'm not sure we should have anything specific here
I'm not quite sure what you mean -- does "I'm not sure we should have anything specific here" mean you think we should skip carbon dioxide, or lump it together with carbonyls (by generalizing the SMIRKS for carbonyls)?
I'm still of mixed opinions here. The idea of adding specialized parameters to cover what can only over be a single molecule does not appeal to me very much. So if we cover it, I'm inclined to say we should lump it together with more general parameters. But I could be convinced to just not cover it.
We discussed this in person, I think it makes sense to skip carbon monoxide and make the first carbonyl group more generic with [#6:1]=[#8X1:2]
I'll make sure it is the first carbonyl coverage so that it doesn't over write anything.
carbon monoxide C#O is being left out, others added in pull request #43
The only bond getting a generic (in my current pull request) on the filtered drugbank set is a C=O in carbon dioxide.
All of the C=O bonds specific
[#6X3]
right now. Carbon dioxide is an odd corner case, I'm not sure if we should actually cover this.