Just one correction. NA names are different (instead of * now there are ', as pdb v3.x requires) but values should be same.
Garib On 17 Mar 2011, at 14:01, Paul Emsley wrote: > On 17/03/11 12:55, Garib N Murshudov wrote: >> Upgrading dictionary should be just matter of downloading from this site: >> >> www.ysbl.york.ac.uk/refmac/data/refmac_experimental/refmac_dictionary_v5.6.0023.tar.gz >> >> or from the latest ccp4 (6.1.24 or something like that) >> >> Garib >> >> On 17 Mar 2011, at 12:54, Paul Emsley wrote: >> >>> On 16/03/11 18:45, Dr. Mark Mayer wrote: >>>> Dear all, >>>> >>>> I'm rebuilding structure refined using phenix -1.7-650 using coot >>>> 0.6.2-pre-1-250. >>>> When I use the rotamers command to select new rotamers (flagged as >>>> bad by molprobity) some of the riding H atoms get left behind, which >>>> means the file is no good for further crystallographic refinement. >>>> Likewise, real space refinement produces the infamous flying hydrogens >>>> problem. >>>> >>>> Any ideas on fixing this? Our current work around is to remove H and >>>> then run reduce again prior to refinement in phenix. >>>> >>>> >>> >>> This is not a Coot problem. There is a mismatch between the model and the >>> dictionary (presumably version 3.x and 2.3). >>> >>> The dictionary needs to be upgraded to 3.x - it is not hard work but needs >>> to be done. I have scheduled it for 0.7 if it is not done before I do it - >>> for standard protein residues, RNA and DNA at least. >>> >>> You are using the approved work-around. >>> >>> Paul. > > Sadly not as easy as that. > > First, Coot uses the "old-style" filenames - i.e. the .cif files [1] are in > lettered subdirectories, not all files in one directory (straightforward for > me to fix if you mean to keep the new naming style). > > Secondly (and more importantly), the PDB version 3.x has a " H " hydrogen > atom on the peptide N - so did the old-style dictionary. The new Refmac > dictionary does not (it does have HN1 and HN2 - presumably non-polymer) - > hence we still get flying hydrogens - different ones :-/. > > So, do I take it that with the new dictionary I now need to apply the > mon_lib_list's data_mod_list items for the peptides (in this case DEL-NH1 and > DEL-OXT) when Coot finds a peptide link? It seems to me that if this > dictionary is out in the wild I'll need to sort this out pretty sharpish. > > Paul. > > [1] the Corina manual calls CIF files "CCP4 dictionary files" :-)
