Hi Filip,
I am currently refining a 1.75A structure and wanted to use the riding hydrogen model in pehnix.refine. From what I found in the manual and previous discussions on the mailing list and if I understand well, using riding Hs is advised for every resolution andthe way of doing this in phenix.refine is not Refmac-like but instead I should 1) put Hs using ReadySet, 2) select the ‘riding’ option for Hs and 3) use the PDB containing the H coordinates for deposition, even though they were not individually refined. In this case their contribution to scattering is also taken into account (which makes sense), which is a different way of treating riding Hs compared to Refmac, if I understand well.
I don't know how Refmac handles H so I cannot comment on this; the rest I think is more or less accurate. Note, "riding" option is used by default (unless it is neutron data). Also, for some more details and illustration on how H contribute to overall scattering and R-factors in different cases have a look at "On contribution of hydrogen atoms to X-ray scattering" article here: https://www.phenix-online.org/newsletter/
Would it be possible to comment on this and in case I got things wrong, correct?
Perhaps one more thing we did not mention yet is that it's best to use H towards the end when the model of the crystal structure is reasonably good.
In fact for my structure, afteradding the Hs prior to refinement using ReadySet and using the riding option, I actually got Rfree dropping by 1% (from 19.5 to 18.5) and better clash score (~0.5 according to Molprobity).
This is well expected and in line with the illustrations in the article I referred to above. All the best, Pavel