Dear Pavel, On Wed, Sep 15, 2010 at 07:57:09AM -0700, Pavel Afonine wrote: > Hello, > > a few points to balance the discussion: Your points sound more sound more like a summary than a contribution to the discussion which might confuse inexperienced readers of this thread, especially if they did not follow it completely. So here me counter-balance ;-)
> > - if you refined your structure without H then it's obviously ok to > deposit it without H; I do not disagree but would like to add that I believe riding atoms usually improve the refinement even at poor resolution, so one should not refine without them at the final stage in the first place. > > - if you refined your structure with H, then you should deposit it with > H, as your refinement software outputs it (if your software uses H but > removes it automatically for you - then it least it's not your > responsibility). Any post-refinement manipulation on final refined model > is bad since it tends to invalidate the reported statistics (R-factors, > for instance), which is illustrated in this paper (section 3.1.5: J. > Appl. Cryst. (2010). 43, 669-676). Indeed, let's not add more > inconsistencies to the database because of a fear that insufficiently > trained people may misinterpret it. I disagree, because since the H-atom are (usually) in a riding position and used e.g. for anti-bumping restraints, they should be considered as (software dependent, as George pointed out) restraints rather than the actual model in terms of coordinates. > > - removing H, if really needed, is a matter of one trivial command, but > adding them back the exact same way they were originally is less > straightforward. I despise the word 'trivial' and as much as there is a 'useless use of cat' there is probably also an 'unnecessary use of trivial'. Cheers, Tim > > - I agree that the X-H distances used in refinement and in validation > are slightly different, although I'm not sure how much of difference > that would make for validation. > > Pavel. > > > On 9/14/10 10:38 PM, Ed Pozharski wrote: >> Mark, >> >> On Tue, 2010-09-14 at 13:34 -0400, Dr. Mark Mayer wrote: >>> Where does the crystallographic community stand >>> on deposition of coordinates with riding >>> hydrogens? >> Surely community is divided on this. There could be arguments made both >> ways. Personally, I think that riding hydrogens can be calculated if >> necessary using the same algorithms/parameters employed upon refinement. >> It is true that different programs may use different parameter sets and >> reproducing exactly the same set of riding hydrogens may be difficult >> without exact knowledge of which version was used and ability to unearth >> that old version of the software. This may preclude one from getting >> exactly the same riding hydrogen positions (how large that difference >> would be I honestly don't know). But really, who cares? What is the >> benefit of knowing exactly where this or that riding hydrogen was? >> Maybe there is some benefit of such comparison in method development, >> but I would think its rather limited. >> >> I wholeheartedly agree with Ethan (even though that is not strictly what >> he said :) that some minor benefit here is completely negated by the >> danger of perception that somehow models tell us where hydrogens are. >> It is bad enough that, in my estimate, roughly 10% of atomic coordinates >> in the PDB are unwarranted as they come from disordered residues with >> exact spatial positions unsupported by electron density. Let's not add >> more things that PDB users may over-interpret. >> >> Cheers, >> >> Ed. -- -- Tim Gruene Institut fuer anorganische Chemie Tammannstr. 4 D-37077 Goettingen GPG Key ID = A46BEE1A
signature.asc
Description: Digital signature
