*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
Dear all,
let's see if I can make a smooth transition of the bulletin board to
a different topic.
Marc Schiltz mentioned that H atoms are not routinely refined because
more often than not the resolution of the data does not warrant their
refinement - except I think there are a few H atoms in proteins whose positions
_are_ uniquely defined by the position of the atom thy are bonded to, and the
atoms bonded to that e.g. the amide Hs, and the Hs on the Calpha, Asn and Gln
amide Hs, Proline Hs, and so on. In fact, most Hs are a pretty sure bet except
the Ser, Tyr and Thr -OH moieties, the -SH of Cys, the terminal -CH3 of Ala,
Met, Ile, Val, Leu, Thr, and the ones bound to N on Arg, Lys, His side chains.
Some stereochemical guesswork could be done on those but I do not want to open
a can of worms there.
Anyway, those H atoms could be added at positions that are essentially
correct within the typical resolution of the data, at zero parameter cost. They
scatter, there are quite a few of them, and they would actually provide extra
non-bonded restraints given that the H-X bond has a direction and a length (for
example MolProbity uses H atom addition to detect clashes pointing to
suboptimal parts of the model). This is a case of a very reliable prior which
we should incorporate all the more in our model, given that the data are
uninformative as to that part of the model.
This is already done for example in Shelx with riding H atoms - so an
option that would "add safely locatable Hydrogens" would help all refinement
programs.
As the Morris cars fans magazine's title goes: "Minor Matters"
Ciao
Pietro
