Hi Dmitry,
COOT does have the MAN-SER linkage record in its monomer lib, but it won't
detect the bond for you. It also haven't provided an interface for the user
to specify the bond type yet.
The COOT procedure you described is perfectly fine for generating a generic
covalent bond record for a PDB file. However without specifying the linkage
to be "MAN-SER" you do lose some restraints for the bond angles (as Engin
pointed out in his post regarding phenix.refine).
It seems that refmac is the only program (that I've used so far) that will
automatically detect the N- and O-glycan linkages and put that information
into the header of a PDB file. This seems to be the easiest way of
generating a proper "LINKR....MAN-SER" line in the PDB header. Note that
refmac writes a LINKR record for its own use, not "LINK" which is the PDB
standard. You can simply manually change the "LINKR" to "LINK " to turn it
into a PDB standard record, but LINKR works fine for newer versions of COOT
too. The advantage of taking this path instead of writing a LINK line
manually (or modifying the LINK line you get from COOT) is that you can be
sure that the "MAN-SER" is put in the right columns - PDB format is
position-sensitive.
Alternatively, here is a LINK line that you can modify to fit your sequence
and paste to your PDB file. Just make sure you only replace characters, not
inserting/deleting any:
LINK C1 MAN B1001 OG SER B 912
MAN-SER
For phenix.refine, yes, it does not write any linkage information to the
header or use those information stored in PDB headers. But you can always
paste that LINK line you get from other places to the PDB files generated by
phenix.refine. There is no need to redo the building in COOT every time.
The LINK line only specifies the bonding between two atoms. So as long as
you haven't renumbered the residues or changed the chain IDs, you can
transfer the same LINK line among your PDB files of the same protein (i.e.
you do not even have to remake the link in COOT for every structure of the
same protein: just paste the LINK line and make sure the residue numbers are
right).
Zhijie
-----Original Message-----
From: Dmitry Rodionov
Sent: Thursday, November 21, 2013 4:09 PM
To: CCP4BB@JISCMAIL.AC.UK
Subject: Re: [ccp4bb] Dealnig with O-linked mannose
Thank you all for helpful suggestions.
My question was how to properly connect a mannose to a serine and real-space
refine the result. My apologies for not being clear enough.
Coot can't find MAN-SER or SER-MAN in it's library (Coot 0.7.1, mon_lib
5.41)
It does not automatically make the bond between MAN C1 and OG of SER either.
Here is the way I finally made the connection followed by refinement it in
Coot:
1) Get monomer... MAN
2) real-space refine MAN into reasonable position
3) Delete hydrogens and reducing hydroxyl
Bond is not detected
4) Extensions->Modelling ...->make Link (Click 2 atoms) ... Click on C1 of
MAN and OG of SER
dashed bond appears
5) Sphere refine
A whole different issue that was touched in the replies is how this model
will be handled by refinement programs.
For the sake of a record, I am using Phenix and it seems to not respect the
described link without massaging by means of either pdb.edits or
apply_link.def.
Also, phenix.refine does not produce LINK records in the output PDB, so step
4 might have to be repeated.
Best regards,
Dmitry
