This is somewhat a tricky question: when refining ephemerous structural intermediates, should we be more tolerant with geometry outliers or not? The background:
I’m getting a large number of different maps from 3DVA (clusters mode). Some of them might represent states that are not populated during a long time (conclusion from biochemistry experiments, although I am not able to tell exactly how long they last, individually). Those are at 3 - 3.5 ish angstrom resolution based on the FSC 0.143 criterium, which means maps do not look great all around, but I am wondering if the current validation standards of the PDB have the same power telling right and wrong for these structures in particular (concerning Rama / rota outliers, planarity, bond angles and lengths etc). Any insights? I’m looking for any reading / watching suggestions, specially those based on very high resolution maps, if possible. This is more like a curiosity for me for now, but it might be useful in a Discussion section, or in a review in the future…
If anything one should be less tolerant of outliers at lower resolution - because you have less justification for them. Ideally (assuming you have a good starting model), you would refine with strong enough restraints (e.g. local reference restraints a la proSMART or REFMAC jelly body) that geometric outliers do not persist. In practice of course, this can be a little tricky to get right, but I definitely wouldn’t use lower res/poorer maps as a justification for including outliers
No, that is not my point. Even in crystallography I am used to resolutions worse than 3 angstroms, and you are totally right, we employ all the tools to reinforce geometry for those. My question is specific to maps coming out from flexibility analysis, which might include intermediates that are not energetically favorable - and that the reasons for that instability probably/possibly include bad geometry. Obviously we do not measure how long they last, so this is a tricky issue, I think.
Maybe - but unless you have justification from either the density (at higher resolution) or some other source of info (molecular dynamics maybe?) I don’t think it is wise to deliberately include outliers… in terms of structural studies of energetically unfavorable intermediates, perhaps some of the literature from time resolved crystallography at XFELs may be helpful - don’t have a good reference off the top of my head though, maybe others will be more help
Sure. I’m afraid we are just opening a window to a completely new universe. Even in time-resolved/Xfel, the objects are crystalline, so energetically favored. The question on CryoEM then, is how many particles one can expect to observe for a conformation that is unfavored. Maybe they are so fast that my supposition is complete nonsense. I agree, for the moment this is mostly a question for molecular dynamics/quantum mechanics people. I’ll keep looking.