Dear all, we also have a lot of "fun" with LINK records when taking deposited PDB structures into BUSTER refinement. Sometimes (but not often) there are missing LINK records, but the vast number of problems is due to erroneous LINK records added by annotation software it seems (based on a cut-off of 3.5A apparently).
Correcting those issues once reported is one option, but maybe revisiting the idea about auto-generating LINK records during deposition would be beneficial. At the end of refinement the various packages will have all required and no wrong/additional LINK records in the final PDB file - and deposition software should probably take those as-is without automatically rewriting/adding/removing any (while still present a note/warning to depositor about potentially missed LINK records according to annotation software). This would most likely result in much fewer LINK problems in deposited PDB structures. As the format guide says: The LINK records specify connectivity between residues that is not implied by the primary structure. and specifically mentions bonds (not hydrogen bonds or salt-bridges): https://www.wwpdb.org/documentation/file-format-content/format33/sect6.html#LINK Cheers Clemens On Sun, Nov 11, 2018 at 07:31:55PM +0000, Robbie Joosten wrote: > Erroneous LINK records happen quite a lot and used to be the combination of > aggressive annotation software and depositors not paying attention to the > comments from the annotators. They make up a large fraction of the bug > reports I have sent to the PDB over the years. They are usually fixed very > quickly by the annotators, as long as someone takes time to report them. > > This case looks like an error in a refinement program which nevertheless > should have been caught by the depositors. What I would like to know is > whether the deposited, pre-annotation model had the LINKs or not. > > LINKs are a bloody nightmare when it comes to annotation. At the moment there > is no record keeping of targets and chemical modifications in a dictionary on > the side of the PDB so there is also no standardisation. IMO mmCIF makes it > easier to store the restraints with the coordinates, but there is still no > neat mapping by LINK identfiers the way the LINKR format works in Refmac. I > think that is a missed opportunity. > > Sorry for the rant, I blame the F1. > > Cheers, > Robbie > > > Op 11 nov. 2018 19:47 schreef Tristan Croll <[email protected]>: > > I've seen instances like the following in roughly half a dozen deposited > structures over the past year or so. Each time I've contacted the > authors, who've been just as mystified as me by them - and certainly > didn't add them on purpose. It seems to me that some fairly > commonly-used package is erroneously turning clashes into LINK cards in > some circumstances. I just found the following clearly wrong LINKs in > 6caj (deposited January this year): > > LINK CD2 PHE I 266 CG2 THR I 272 1555 1555 > 1.47 > LINK CE2 PHE I 266 CG2 THR I 272 1555 1555 > 1.47 > > ... which looks like the attached image. The same bonds are also > specified in the mmCIF file, for the record. > > Anyone have any clue? > > Best regards, > > Tristan > > ######################################################################## > > To unsubscribe from the CCP4BB list, click the following link: > https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB&A=1 > > > ######################################################################## > > To unsubscribe from the CCP4BB list, click the following link: > https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB&A=1 -- *-------------------------------------------------------------- * Clemens Vonrhein, Ph.D. vonrhein AT GlobalPhasing DOT com * Global Phasing Ltd., Sheraton House, Castle Park * Cambridge CB3 0AX, UK www.globalphasing.com *-------------------------------------------------------------- ######################################################################## To unsubscribe from the CCP4BB list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB&A=1
