Closed electronicsbyjulie closed 1 year ago
The IDs of the removed PDBs are:
2ydo 3rfm 5g53 2rh1 4ldo P38405
It is downloading the PDBs, opening the receptors and alpha subunits, uprighting the assemblies, and outputing them beautifully to example coupled PDBs.
Next step will be to examine the mTAAR9-hGNAS2 examples having different ligands, and see how much they vary in order to estimate whether "soft" docks will be necessary. Soft docking is far more CPU intensive than normal docking. A small variance between models may mean that soft docking would not be important, but ORs have "softer" binding pockets than TAARs.
Then there will have to be some kind of layer between the coupled PDBs and the PrimaryDock app, or else a new feature internal to the app, to do the conformational change of the OR in parallel to the coupled PDB and its inactive counterpart. This functionality will first apply the changes to activate the inactive OR for a given G protein, then fine-tune the receptor's shape to optimize G protein contact, optimize internal contacts such as Y6.55-D/E45.51, and minimize clashes. Finally, the custom tuned receptor will be ready for dock.
The results are looking great!
Recall that we want to create coupled versions of the odor receptor PDBs and store them in the pdbs/coupled/
folder, then do docking against those. The idea is for the coupled models to be reusable for all ligands. Therefore, they wouldn't be generated on the fly by a function of bin/primarydock
.
This is almost ready, just have to figure out how best to fit in reshaping the receptors to minimize internal clashes and clashes with the G protein.
It's making promising looking predictions so far. I'm thinking of readying this for merge and starting on a UI feature that would display predictions on the odorant page in the web app.
Minimizing clashes is more important than the UI, and the receptor conformations should be fixed in this PR.
The UI feature can wait until the server has processed at least a few hundred receptor-odorant pairs.
This PR has gotten way too big. Let's make sure it passes the unit test and then ready it for merge. @ssepeq
Still want to get couple.pepd
to avoid clashes between the helices and the G protein.
Also there are several internal contact residue pairs that should be reestablished after homology conform.
1.60 - 7.66
2.39 - 3.46
2.42 - 3.42
2.53 - 3.32
3.30 - 4.54
3.36 - 6.47
3.40 - 6.48
3.56 - 5.60
45.47 - 6.62
45.49 - 7.41
Once the clash avoidance is working, go ahead and test and mark the PR ready for review. The contact residues can be addressed in a new ticket.
Created #292.
Hoping ticket 292 fixes problems with residue 292 of OR51E2 reaching the ligand. 🤡
Test succeeded.
Not wanting to wait for the requested changes on this ticket, the 3 letter fix has been added to stable
so that the current stable branch is no longer broken.
PR rejected.
This is mutually exclusive with #286 and #289 as there would be many conflicts due to copied code.
Trying something different. If we can do homologous conformational changes to ORs based on the OR51E2 cryo-EM, then why not for the mTAAR9 cryo-EMs and various other experimental models of coupled GPCRs. The placement of the G protein can be part of the homology model.