addh wrong hydrogen name on glycosylated Asn residues

[Elaine Meng]

When the asparagine is glycosylated, since there is only one hydrogen added to the sidechain N it gets named HD2, whereas the template hydrogen names are HD21 and HD22. An example structure with hydrogens on glycosylated Asn is 1ah1 (has name HD21). An example for testing addh with both glycosylated and unglycosylated asparagines is 5o4o.

There may be analogous situations but I can't think of any off the top of my head. Serines and tyrosines may be O-glycosylated, but then addh won't add any hydrogens to that oxygen anyway.

Daniel Asarnow reports that the name HD2 causes problems in Phenix.
<​http://www.rbvi.ucsf.edu/pipermail/chimerax-users/2020-April/001073.html>

[Tristan Croll]

It's not just that - the hydrogens on the sugar itself (and on any other 
ligands) will mostly have names that don't match the CCD definition. I 
have a fix_residue_from_template() method in ISOLDE designed to correct 
a given residue against its CCD template by correcting the atom names 
and filling in any missing atoms - I've been meaning to add an optional 
argument to it to just rename the atoms, since for some residues it adds 
atoms that actually *shouldn't* be there. I could polish that up and 
turn it into an ISOLDE command, or donate it to ChimeraX if you like.

On 2020-04-24 17:22, ChimeraX wrote:

[asarnow@…]

Just want to confirm, editing to HD21 results in the expected behavior.

Best,
-da

On Fri, Apr 24, 2020 at 10:23 AM ChimeraX <ChimeraX-bugs-admin@cgl.ucsf.edu>
wrote:

[pett]

If I compare the names that ChimeraX uses when adding hydrogens to NAG (e.g. /A:1547 in 6jx7) to those from an NMR structure that already has hydrogens (e.g. /A:132 in 1ah1), they are the same. Perhaps ChimeraX goes astray when the structure geometry is so bad that it adds the wrong number of hydrogens to various NAG atoms?

[Tristan Croll]

Oh - sorry. Good to know the sugar hydrogen names do come out right. That’s what comes of making assumptions... this does become a problem in general with many ligands, though. 
 

 

[asarnow@…]

Just another piece of information, in my case the NAG and ASN here are
simulated successfully in Isolde.

-da

On Fri, Apr 24, 2020 at 10:18 PM ChimeraX <ChimeraX-bugs-admin@cgl.ucsf.edu>
wrote:

[Tristan Croll]

Yeah - ISOLDE (well... OpenMM, really) doesn’t actually care about atom names. Atoms are matched to the MD residue template according to the bond topology (which is good, because different MD forcefields do wildly different things with atom names).
 

 

[pett]

Fixed in the next daily build. Might put an AddH 1.0.1 bundle on the Toolshed for the 1.0 release.