[phenixbb] Riding H
jzhu at idi.harvard.edu
Fri Oct 29 16:55:07 PDT 2010
As I said, stripping the H and re-add the H has a little problem. Because the methyl group can rotate, phenix.refine actually rotates the H atoms on the methyl group, a lot of clashes can be avoided. While the molprobity and phenix validation tools can't do that, a lot more clashes appear.
On Oct 29, 2010, at 6:35 PM, Nathaniel Echols wrote:
> On Fri, Oct 29, 2010 at 2:56 PM, Jianghai Zhu <jzhu at idi.harvard.edu> wrote:
>> That is what I thought. But wouldn't the nuclear distance be more accurate? and we should use the molprobity bond length, the longer one?
> I'm not sure what is most accurate from the standpoint of a molecular
> modeler or chemist, but as far as X-ray diffraction is concerned, the
> shorter distance is better - however, the difference in R-factors is
> minimal, even at atomic resolution. Internally, we're using the same
> monomer library that Refmac uses, so it's non-trivial to switch to the
> longer distances. I suspect it might improve the geometry, but I
> haven't tested this.
> For what it's worth, the validation tools in PHENIX (which includes
> most of Molprobity at this point) always strip and re-add hydrogens
> before calculating clashes, so the result will always reflect the
> longer bonds.
> phenixbb mailing list
> phenixbb at phenix-online.org
More information about the phenixbb