Dear Herman,

> Some time ago, I had similar experiences with attached sugars. My
> impression was that the added LINK cards were ignored unless the ND2-C1
> distance in the model was very close to the ideal distance. Due to the
> automatic determination of chirality, I ended up for some sugars with
> the wrong chirality, which I still have to correct. The idea of using a
> well-refined glycan as a template is a good one, which I definitively
> will try.
In (recent versions of) Refmac the LINK records do their job iff you
choose the right settings. I admit it is not the default, but 'make sugar
NO' works fine for me as long as the LINK(R) records are correct.

> Fact is that attached sugars generally have multiple conformations and
> significant disorder and by just fitting them to the available blobs in
> density it is easy to end up with wrong conformations.
It is easy to get things wrong, I agree, but that is also true for some
parts of the protein. We still try to build them. When the disorder is too
bad you probably don't want to build that part of the structure. Discrete
multiple conformations (haven't seen many for carbs) can be built and
refined if the restraints are chosen with care.

> In addition, the
> sugars are in general completely irrelevant to the medical or biological
> question one wants to solve with the crystal structure, so one would
> like to waste as little time as possible with the fitting and refinement
> of those sugars.
Exageration: "Wasting time is never good, that is why I only build the
active site of proteins and leave the rest of the structure to the
imagination of others."
It would be pretty scary if people would deposit such structures. The fact
is that carbs may not answer YOUR question, they may very well answer the
research questions of other users of your structure. That is a key point
of depositing structures: allowing others to use them for their research.
Without wasting resources to solve the structure again.

Carbohydrates play important roles in protein folding, multimerisation and
stability. The particular nastiness of the Spanish flu was also linked to
the carbohydrates in neuraminidase (Wu ZL, Ethen C, Hickey GE, Jiang W
(2009) Active 1918 pandemic flu viral neuraminidase has distinct N-glycan
profile and is resistant to trypsin digestion. Biochem Biophys Res Commun
379(3):749–753). Carbs can be very interesting for some of us.

> For this, we need robust definitions of the geometry of the sugars and
> the links to the protein. This could be in the form of special residues
> like BMA, or by specific LINK cards. Important is that they enforce the
> correct geometry, even if the fitted conformation is far off from what
> it should be.
These defenitions already exist. One should not expect Refmac to do magic.
It is a refinement program which tries to optimise stuff we build. This is
understandably garbage-in/garbage-out.

That said, you make a very clear case for the improvement of current
carbohydrate building algoriths. There has already been a lot of
improvement in that field lately, so I'm sure we'll get there.

Best wishes,
Robbie Joosten

>
> Best regards,
> Herman
>
> -----Original Message-----
> From: CCP4 bulletin board [mailto:[log in to unmask]] On Behalf Of
> Damian Ekiert
> Sent: Wednesday, December 23, 2009 2:52 AM
> To: [log in to unmask]
> Subject: Re: [ccp4bb] Coot pudding? (a.k.a N-linked carbohydrate
> addition)
>
> Pierre,
>
> For what its worth, I just tried using "Real Space Refine Zone" on
> something I am working on now (2.8A resolution, in coot 0.5.2, build
> 1691). With the "LINKR" record generated by REFMAC5, my glycan slithers
> down into the protein density. Similarly, using "Regularize Zone" does
> not enforce the Asn-NAG linkage. I also tried replacing the "R" from
> LINKR with a space, but it doesn't make a difference.
>
> My records read:
>
> 1 2 3 4 5 6 7
> 8
> 123456789012345678901234567890123456789012345678901234567890123456789012
> 34567890
>
> LINKR C1 NAG A 401 ND2 ASN A 21
> NAG-ASN
> LINK C1 NAG A 401 ND2 ASN A 21
> NAG-ASN
>
> However, I checked the page that Paul mentioned in an earlier email, and
> it states:
>
> "LINK and LNKR cards are not used to determine the geometry of the
> restraints."
>
> If other people don't have this problem, I would love to know how to get
> this to work because, admittedly, placing everything using other tools
> is a pain.
>
> Best,
>
> Damian
>
>
>
>
>
> Pierre Rizkallah wrote:
> > Hi Damian,
> >
> > I have used both Coot and Refmac for glycan modelling/refinement in
> the past, last time was 2 or 3 years ago. They both behaved correctly
> for me.
> >
> > To get the correct behaviour, you must have the correct LINK cards in
> the header. At the time, the Chemical ID BMA did not exist in the PDB.
> There was only MAN. Presumably, the PDB had originally decided that the
> saccahride linkage is a bit like the peptide bond, where it can be one
> sort or another, so you have to specify it explicitly. But convenience,
> or avoiding confusion, must have been the reason for the recent creation
> of BMA. If Coot/Refmac do not have it in their library, well you can add
> it easily into the correct folder on your disk, by editing the
> MAN-b-D.cif entry and calling it BMA.cif. Garib told me, when I was
> grappling with this problem, that Refmac works out from the coordinates
> which form of the saccahride linkage it is and imposes the correct
> restraints. Admittedly, there must be room for confusion if modelling is
> not accurate enough. So BMA is better all round.
> >
> > The branching of the glycan tree does have a good template in PDB
> entry 1LGB, first deposited in 1994. It has been reviewed to include BMA
> earlier this year. The branching pattern is correct, so care must be
> taken to enter the correct numbers in the LINK cards if you are to
> reproduce a similar branching pattern in your glycan. Incidentally, if
> you search the PDB for chemical id BMA, there are apparently 36 entries.
> 1LGB was not among them, and I don't know why.
> >
> > I hope this is useful. Good Luck.
> >
> > Pierre Rizkallah
> >
> > ********************************************************
> > Dr. Pierre Rizkallah, Senior Lecturer in Structural Biology, WHRI,
> > School of Medicine, Cardiff University, Heath Park, CF14 4XN
> > email: [log in to unmask] phone + 44 29 2074 2248
> >>>> Damian Ekiert 22/12/09 5:46 PM >>>
> > For an "ideal" glycan, you could used a model from a high resolution
> > structure, or something that has been energy minimized, etc. Mostly I
>
> > find this helps in getting the sugars in about the right place
> > (keeping bond lengths and angles reasonable).
> >
> > I perhaps could stand to fiddle more and maybe I'll look through the
> > updated documentation. Last I tried, the Asn-NAG1 linkage wasn't
> > enforced, with would allow the whole glycan to slide down into the
> > protein, or off into space. I am typically building relatively small
> > glycans (2-5 residues) into ~3A data, so the density itself doesn't
> > keep things in place very well.
> >
> > Regarding BMA vs MAN: When I have tried to used BMA in REFMAC, it
> > doesn't seem to recognize it and requires a library file. But if you
> > use MAN, it adds a MODRES record to the header, enforcing beta-mannose
>
> > geometry. Not sure if this is just a REFMAC version issue or what.
> >
> > Best,
> >
> > Damian
> >
> >
> >
> >
> > Paul Emsley wrote:
> >> If I can chip into this somewhat sacrilegiously-named thread
> >>
> >> 1) I *would* use real-space refinement :), specifically Sphere
> >> Refinement. You can dial down the
> >> density weight if needed, of course.
> >>
> >> 2) the documentation on refining carbohydrates in Coot has recently
> >> been updated
> >>
> >> http://www.biop.ox.ac.uk/coot/doc/coot/Refining-Carbohydrates.html
> >>
> >> 3) Coot does not (yet) correct chiral centre inversions in glycosidic
>
> >> linkages on refinement
> >>
> >> 4) or delete the O1s :)
> >>
> >> Paul.
> >>
> >>
> >>
> >> Robbie Joosten wrote:
> >>
> >>> Dear Steve,
> >>>
> >>> I would also use Damian's approach, but the sequence of the core
> >>> should be NAG-NAG-BMA-(MAN)2. This is improtant because the correct
> >>> stereochemistry restraints for beta-mannose can only be applied when
>
> >>> you call the residue BMA.
> >>> Building carbohydrates also comes with special validation
> requirements.
> >>> PDB-care and CARP are both very usefull. Unfortunately, the service
> >>> is currently down (http://www.dkfz.de/spec/glycosciences.de). Just
> >>> make sure the links between your carbs are correct and, please,
> >>> remove the O1 atoms when needed ;)
> >>>
> >>> Cheers,
> >>> Robbie Joosten
> >>>
> >>> ----------------------------------------
> >>>
> >>>
> >>>> Date: Mon, 21 Dec 2009 17:48:31 -0800
> >>>> From: [log in to unmask]
> >>>> Subject: Re: [ccp4bb] Coot pudding? (a.k.a N-linked carbohydrate
> >>>> addition)
> >>>> To: [log in to unmask]
> >>>>
> >>>> Steve,
> >>>>
> >>>> My general strategy is to start with an "ideal" glycan (an Asn
> >>>> linked to
> >>>> NAG-NAG-(MAN)3 ) and superimpose the Asn on the residue from my
> protein.
> >>>> Then you can move the whole glycan as a rigid body until the Asn
> >>>> and first NAG are roughly positioned. Then you can tweak any sugars
>
> >>>> further out on the chain to get them to fit. Unless you have really
>
> >>>> great density, usually it is best to avoid real pace refine zone.
> >>>> Better to fit the sugars using the manual rigid body fitting tools,
>
> >>>> do the best you can, then REFMAC usually brings them in OK.
> >>>>
> >>>> I have some models that I could send you if you need them.
> >>>>
> >>>> Best,
> >>>>
> >>>> Damian Ekiert
> >>>>
> >>>>
> >>>>
> >>>> Soisson, Stephen M wrote:
> >>>>
> >>>>
> >>>>> Hi everyone-
> >>>>>
> >>>>> I was searching for some information on what might be the best way
>
> >>>>> to add N-linked sugars in coot, and Google has let me down.
> >>>>> Searching "adding sugars in coot" returns a very nice recipe for
> Coot Pudding.
> >>>>>
> >>>>> ***_Recipe for_/ Coot//__/_ Pudding - American_/ Coots/*
> >>>>> ******** It has plenty of fat,/
> >>>>> sugar/, and starch, and probably some calcium from the milk.* ...*
>
> >>>>> The/ coots/ will not tolerate/ adding/ eggs in any form, so this
> >>>>> is an egg*
> >>>>>
> >>>>>
> >>> ...*
> >>>
> >>>
> >>>>> ///_www.beaky_//_*coot*.com/pudding.html_///
> >>>>> ///// -/ _Similar_
> >>>>> //
> >>>>>
> >>>>>
> >>>>>
> >>>>> I did not know that coots had such an aversion to eggs. :)
> >>>>>
> >>>>> Anyway, would anyone have any top tips on adding N-linked sugars
> >>>>> using coot? I can import the NAG monomers, but linking them up to
> >>>>> the protein seems non-trivial
> >>>>>
> >>>>> Many thanks in advance,
> >>>>>
> >>>>> Steve
> >>>>>
> >>>>>
> >>>>> Notice: This e-mail message, together with any attachments,
> >>>>> contains
> >>>>>
> >>>>>
> >>> information of Merck & Co., Inc. (One Merck Drive, Whitehouse
> >>> Station, New Jersey, USA 08889), and/or its affiliates Direct
> >>> contact information for affiliates is available at
> >>> http://www.merck.com/contact/contacts.html) that may be
> >>> confidential, proprietary copyrighted and/or legally privileged. It
> >>> is intended solely for the use of the individual or entity named on
> this message.
> >>> If you are not the intended recipient, and have received this
> >>> message in error, please notify us immediately by reply e-mail and
> >>> then delete it from your system.
> >>>
> >>>
> >