@rbs_sci Thankyou so much for your detailed reply 
Id like to respond as a sanity check to each point
First: I think all of these concerns are valid, and these are the issues i mentioned i can think of; while these concern me, i think in some of these cases there are mitigations, for example, using substantial overlap in the areas with quality density to minimize incorrect density offsets.
Second: Yes this is obviously an issue on the map front; in theory you can still model a protein and validate the geometry and model map fit (though this does pose the risk of introducing slight but potentially meaningful geometry errors if there are density offset errors), and individual component maps and concensus map are required as mandatory additional depositions on PDB nowadays so the maps in theory can be independently validated.
Third: Thankyou so much for the paper, i will read this i have to date found that Phenix works best to minimize noise between component maps. For strictly visualization purposes, i have in past used chimeraX delete density tool to remove noisy periphery of component maps, preserving only the high resolution overlapping regions. This produces a composite density that looks far less noisy. I want to stress i have never deposited one of these maps as deleting parts of density even if its done rationally to remove noise seems dodgy. Do you agree this should not be deposited?
Fourth: this leads on a bit from the last comment. I agree this is a big problem; i have found it far less of a problem in phenix, as phenix essentially crops the best map around each model, meaning there is no noise added from the maps that are poor quality at those locations. Also i am working at 2.8-3.5 Ă… so i never model waters, and am careful with what i do model. The densities are very clean for this resolution so protein is pretty unambiguous. Also i normally use overlap on a case by case basis, but inspect the result to determine the quality of the overlap region.
Fifth: absolutely definitely i am! however, this is the only way to achieve high resolution of the region and does represent what i believe is the only stable conformation. The structure features a hinging mechanism, that appears to have a free range of motion when not “locked” so i don’t think it is possible for me to see the other conformations, however i can view the “locked” conformation as a composite map at high res which is better than no structure i figure.
Finally, i have deposited composite maps before i was informed of how critical people are of them. I would always deposit the maps with clear title indication of the fact that they are composite, all component maps linked, and the concensus map. The models is the difficult part i feel as i said some proteins cannot be resolved in a single map.
I could in theory generate several protein fragments in each map and deposit them seperately, but i feel this would be more confusing, and that a single model in a composite map composed from high quality overlapping maps is probably better and still fairly structurally accurate despite small introduced artefacts?
Sorry for making this so long, its just a complicated question and there are few people i can ask around me!
James.
) and stitching artefacts can still be a pain.