Hi John and Eleanor, The absence of a LINKR equivalent in the mmCIF dictionary is a shame. The mapping to a full description of the bond (bond order, residue modifications, geometric targets) rather than just a statement that two atoms are connected (with a limited set of types), is something that could really improve the PDB as a databank. This would require and extension of the CCD to also contain descriptions of linkages like the CCP4 dictionary has (in a limited way), but that makes the data more completely annotated.
Cheers, Robbie Sent from my Windows 10 phone ________________________________ From: CCP4 bulletin board <CCP4BB@JISCMAIL.AC.UK> on behalf of John Berrisford <j...@ebi.ac.uk> Sent: Sunday, December 10, 2017 9:28:30 PM To: CCP4BB@JISCMAIL.AC.UK Subject: Re: [ccp4bb] LINK vs LINKR Dear Eleanor mmCIF does not have a equivalent of LINKR records. If you define links during refinement then these are output as _struct_conn in mmCIF files. The _struct_conn records are used during deposition to determine if two atoms are linked, allowing using to validate chirality (such as ASN to a NAG) and ignore these atoms in clashes. Any _struct_conn records which are invalid due to one of the atoms missing will not affect the analysis at deposition, they will just be ignored. Please try outputing an mmCIF file out of refmac and report any incorrect or missing _struct_conn records to the refmac developers. John On 10/12/2017 20:04, Eleanor Dodson wrote: Dear John - sorry for my ignorance but does mmCIF provide a proper LINKR replacement? Surely there is just as likely to be redundant LINK records as LINKR? It is easy enough to get rid of a residue and forgetting to leave useless CISPEP SSBOND LNK etc lines in the header... Eleanor On 10 December 2017 at 19:59, John Berrisford <j...@ebi.ac.uk<mailto:j...@ebi.ac.uk>> wrote: Dear Bernhard When we convert from PDB to mmCIF we do indeed ignore LINKR records as these are also present between unmodelled residues in the PDB file (or were when we tested). However, if you output an mmCIF file directly from refmac then the link records are correctly written out into _struct_conn records (mmCIF equivalent of LINK records) which are correctly handled by the deposition system. To do this simply add the keyword "pdbout format mmcif" to your refmac commands and it will output an mmCIF file ready for deposition. I hope this helps John PDBe On 09/12/2017 20:34, Bernhard Rupp wrote: Dear Developers, the TER issue in REFMAC seems to be fixed, but is really necessary that REFMAC places LINKR records instead of LINK records in the PDB header? The PDB does not recognize those, and although it’s only a minor nuisance to fix them in an editor, one tends to forget this between revisions…or the PDB could simply ignore the ‘R’…. Best, BR ------------------------------------------------------ Bernhard Rupp http://www.hofkristallamt.org/ b...@hofkristallamt.org<mailto:b...@hofkristallamt.org> +1 925 209 7429<tel:%28925%29%20209-7429> +43 767 571 0536<tel:+43%207675%20710536> ------------------------------------------------------ Many plausible ideas vanish at the presence of thought ------------------------------------------------------ -- John Berrisford PDBe European Bioinformatics Institute (EMBL-EBI) European Molecular Biology Laboratory Wellcome Trust Genome Campus Hinxton Cambridge CB10 1SD UK Tel: +44 1223 492529<tel:+44%201223%20492529> http://www.pdbe.org http://www.facebook.com/proteindatabank http://twitter.com/PDBeurope -- John Berrisford PDBe European Bioinformatics Institute (EMBL-EBI) European Molecular Biology Laboratory Wellcome Trust Genome Campus Hinxton Cambridge CB10 1SD UK Tel: +44 1223 492529 http://www.pdbe.org http://www.facebook.com/proteindatabank http://twitter.com/PDBeurope
