Hi Phil, each program handles hydrogen atoms slightly different: - B-factors of hydrogen atoms may be simply inherited from the parent atoms (atom X in X-H bond) or they may be multiplied by a factor 1.0...1.5. These B-factors may be set once at the beginning of refinement run or they may be updated during refinement; - Different X-H bond lengths may be used: shorter distances (corresponding to electron cloud; as described in Shelxl) or longer "neutron distances" (corresponding to nuclear position). Even within one library you can observe some mixtures: for example, in Monomer library shorter distances are used for amino-acids and longer distances are used for some other compounds; - Programs may or may not account for H atom scattering (they may only use hydrogens in geometry restraints). This all makes it not very straightforward to restore hydrogen atoms the same way they were in original structure, which means that the reported statistics may not be exactly reproducible. This is illustrated in our recent paper: Afonine PV, Grosse-Kunstleve RW, Chen VB, Headd JJ, Moriarty NW, Richardson JS, Richardson DC, Urzhumtsev A, Zwart PH, Adams PD: phenix.model_vs_data: a high-level tool for the calculation of crystallographic model and data statistics. J. Appl. Cryst. 2010, 43, 669-67 Finally, removing hydrogens is a matter of a simple command, for example: phenix.reduce model.pdb -trim > model_no_H.pdb while restoring H atoms the exact same way they were originally is much less straightforward. All the best! Pavel. On 8/23/10 1:08 PM, Phil Evans wrote:
Coming from using Refmac I'm slightly puzzled by the philosophy here. By default (I believe) Refmac always adds riding hydrogens for the refinement but doesn't output them in the refined PDB files. This seems to me to be the right thing to do. I don't see any reason for not including Hs in refinement since you know where most of them are relative to the heavy atoms, and they will (maybe marginally) improve things. On the other hand I don't normally want them cluttering up the display in Coot (maybe because I'm just not used to them), and I certainly don't want to include them in my PDB deposition, since I don't think they are an independent part of the model, and including them would imply I can see them. They are I suppose useful in bump checks, but I would just add them back for that purpose
Phil