On 19/06/17 09:54, Chris Ulens wrote:
I am looking to get feedback from others who have recently built and refined bond lengths and angles of an N-linked glycosyl chain in Refmac. I observed in Coot that the bond between O3 of BMA303 and C1 of MAN304 does show while the bond between O6 of BMA303 and C1 of MAN305 does not and I wonder why.
Well, Coot 0.7 went out of date over 4 years ago so it's a bit of a struggle to remember. My guess would be that Coot 0.7 doesn't interpret LINKR records in the way that Refmac does. Using standard rules, 303 and 304 are tandem (and hence are bonded) but 303 and 305 are not (and hence are not bonded).
Molprobity issues warnings for bond length and angle outliers for the entire chain.
Inconsistent dictionaries? What fun. Use Acedrg to generate new ones and as long as your model is consistent with that (both Coot and Refmac can tell you so) then let the pieces fall as they may. The newest version of Acedrg (the binaries for which are not yet distributed by CCP4) can be run in link generation mode which can be used to replace/update the entries for glycosidic linkages in the Refmac monomer library (if you like).
Paul.
