Region clashes

[primaryodors]

This PR adds information about clashes between the ligand and the TM helices to the output of primarydock. It is used by the FYG-active prediction method to determine whether to attempt a per-ligand customized model for the receptor's active state.

[primaryodors]

Currently this is producing false positives for larger SCFA molecules (C>5) in OR51E2.

[primaryodors]

All tests succeeding.

[primaryodors]

The false positives for SCFAs have to do with flexibility of side chains. In the 8F76 model, the side chains of His104, Phe155, and Ile202 stay close enough together to prevent any aliphatic carbons from passing through into a large pocket between TM helices III, V, VI, and VII. But Ile202 is not constrained to the way it's positioned in 8F76, in fact the AlphaFold model has Ile202 positioned in a way that would allow aliphatic chains to reach into the large pocket. The prediction is consistent with what is known about OR51E2's structure, and the reason for the receptor to not be sensitive to SCFAs of more than 5 carbons appears to be something other than the inability of the ligands to fit inside the receptor.

[primaryodors]

⚠️ ❗ The dock is tearing atoms off the side chains of Gln and Glu.

[primaryodors]

The stray atoms issue has not been occurring, however there is now a debug constant to help track it down if it does.

Gln202 is getting mangled in OR56A5.

[primaryodors]

The debug code is not catching the mangling problem of Gln202. Should modify the "Greek continuity" function to ensure that the moves_with array of each Bond object contains all of the residue's higher order Greek lettered atoms.

[primaryodors]

OR56A5:Gln202 seems to be fixed.

[primaryodors]

Tests are succeeding.