OR2AG1 is likely to be misaligned.

[electronicsbyjulie]

Gelis et al (https://doi.org/10.1002/anie.201103980) make a very good case that residues A104, F206, V260, S263, S264, and T279 are involved in ligand binding of n-amyl butyrate in OR2AG1. The article visually depicts molecular docking results from a PDB modelled after a bovine rhodopsin template, and all of these residues are shown in close proximity to the ligand.

Site-directed mutagenesis reveals that changing S263 and S264 to C causes a reduction in activity, consistent with reduced hydrogen bonding, while changing them to V causes a greater reduction in activity, consistent with a loss of hydrogen bonding. Change of T279 to V also causes a loss of function. Changing A104 to G does not significantly decrease activity, but changing it to a bulky I does. Similarly, mutating V260 to a bulky W also decreases activity.

The assignment of secondary structural features is given in the supplementary materials. It shows R269 as the last residue of TMR6. An R or K residue near the top of TMR6 can form a salt bridge with D/E45.51 in the active state, thereby helping stabilize the active configuration. However, the PDB used for the article does not seem to be available.

Here are two views of the binding residues in the AlphaFold model of OR2AG1. They are nowheres near each other and not close enough together to coordinate to a molecule of a fruit ester. R269 (not shown) is placed in TMR7 instead of TMR6, making it less likely to stabilize the active conformation.

If we have to do our own homology modeling, then that is going to make things so much more difficult.

[electronicsbyjulie]

The ZhangLab models resemble the AlphaFold model.

[primaryodors]

It may be possible to install and run AlphaFold locally. https://github.com/deepmind/alphafold

That said, AF models are regarded as being highly accurate, and the article is from back in 2012. Caution should be observed when a site directed mutagenesis study only tests one ligand, as we saw with OR2T11 and copper-mediated thiol sensitivity. Recall that paper predicted a ligand binding site in the cytoplasmic end that conflicts with the G-protein binding site. For OR2AG1, the two serines beyond the extracellular end of TMR6 might actually stabilize the active configuration, while the other identified residues might have functions wholly within the protein.