In Puzzle 2330, I am playing with a ligand that came from the compound library
as one in a group of 4 (the library labels them 3a 3b 3c and 3d). These compounds
all have the same molecular weight and other properties and give similarities of
1.000000 compared to each other. I think each compound contains two chiral carbon
atoms. Each of these carbons has 4 single bonds going to 4 different groups.
One carbon bonds to -COO, -H, a ring, and a long chain. The other
carbon bonds to -CH3, -H, -OH, and a different long chain.
If I compare J3-3a with J3-3b, they differ at one of these carbons.
If I view this carbon with its long chain pointing into the page,
its attached -COO, -H, and ring appear in that order
going clockwise for J3-3a and counterclockwise for J3-3b.
If I view their other carbon with its long chain pointing into the page,
its attached -CH3, -H, and -OH appear in that order
going clockwise for both J3-3a and J3-3b.
Despite the above, I think one of my solutions for J3-3a has changed
from its initial handedness to a different one. Now the carbon with
-CH3, -H, and -OH has them appear in that order
going counterclockwise instead of clockwise.
Is it possible for this to happen within Foldit,
like if you set the clashing importance to 0 and use bands
to pull really hard on the groups around the chiral carbon?
I wonder if this new J3-3a will still give an exact match
for 3a in the library it came from. Maybe it will give an
exact match for 3c or 3d instead.
The chirality of the compounds should be preserved by most operations, but as it's the geometry of the connected groups which determines the chirality (rather than the other way around), if you force the geometry to invert (e.g. by using bands and reducing clashing importance), then the chirality will change.
The compound library might not pick that up, though, as the internal annotation of the compound may be cached from when it was originally generated. (Though if you make any changes to the compound, that may be enough to cause Foldit to re-calculate the chirality based on the current geometry.)
Trying a new chirality is one of the things I'd like to see out of the SMD shake tool. I understand that the Drugit paper mentions it's very limited due to how slow ETKDG computation is, but if and when we can get more conformers I'd like chirality to be in the range of things to tried.
I see the RDKit has a control on whether chirality is enforced, but foldit probably has its non-Python implementation separate from this stuff. https://www.rdkit.org/docs/RDKit_Book.html#conformer-generation
For biochemists, the chirality of a molecule is central to a molecule's identity. (For example, glucose, galactose, and mannose only differ from each other in chirality, but they're distinct enough biochemically speaking to deserve their own names.)
As such, I don't think we'll ever simultaneously sample conformers (which are different positioning of the same compound) with chirality. If conformer sampling ever comes up, it'll likely be a separate tool. (Which might function internally similar to the Shake tool, but it will certainly be labeled differently – though we'll probably start with player-directed chirality changes, rather than automated tools.)
