Thanks to Pavel and Nigel, who pointed out that the LINK records in the pdb file are ignored by Phenix and that I must apply the links with a .phil file in the form:

pdb_interpretation {
    apply_cif_link {
      data_link = "MAN-TRP"
      residue_selection_1 = chain C and resname MAN resid 1004
      residue_selection_2 = chain C and resname TRP resid 269
    }
}

Where MAN-TRP is the name of link description that can be prepared in JLigand. I'm still a little confused since I've done successful crystallographic refinements of both with and without including the apply_cif_link instruction, but this works now.

--
Kevin Jude, PhD (he/him/his)
Structural Biology Research Specialist, Garcia Lab
Howard Hughes Medical Institute
Stanford University School of Medicine
Beckman B177, 279 Campus Drive, Stanford CA 94305


On Mon, Jun 14, 2021 at 12:31 AM Kevin Jude <kjude@stanford.edu> wrote:
I'm trying to refine a cryoEM structure that contains some MAN/C1-TRP/CD1 glycosylations. I've done this successfully in crystallographic reciprocal space refinement, using cif files to define the MAN geometry and the MAN-TRP linkage, plus adding LINK instructions to the pdb file. In this cryoEM structure, though, using the same .cif definition files and similar LINK instruction, the bond is still not defined in the .geo file and instead it's treated as a nonbonded interaction, kicking the mannose away from the indole side chain. The NAG-ASN linkages in the cryoEM structure are fine. Do I need to do anything differently to define these linkages in Real Space Refine compared to phenix.refine?

--
Kevin Jude, PhD (he/him/his)
Structural Biology Research Specialist, Garcia Lab
Howard Hughes Medical Institute
Stanford University School of Medicine
Beckman B177, 279 Campus Drive, Stanford CA 94305