Re: [AMBER] Glycophosphate parametrization using GLYCAM

From: Jochen Heil <>
Date: Thu, 19 Jun 2014 13:57:00 +0200

Hi Francois and Lachele (sorry for the additional "l" in the previous

> You could make yourself a dummy "PO3" residue by substituting atom
> names and types from the SO3 residue.
We actually already tried that. However leap came up with a lot of error
messages concerning missing angle parameters which puzzled us quite a
bit since the corresponding entries were clearly available in the GLYCAM
".dat" file.

> You'd still need to set atom
> types for the attaching carbon (Cp).
And this might be the reason... we'll try.

> After everything is done, you
> still need to assign charges. For this, you can hack the topology
> file if you feel like it. Or, you can create a prep file for your
> molecule as a whole (antechamber) and build a top file from that. If
> you choose the latter, you can have leap write out a pdb from using
> your fake-PO3 prep and base your new prep file on that. You could
> also get a pdb from GLYCAM-Web by building the sulfated analog for
> what you want and change the S's to P's. Then, build a prep file from
> that (and your new charges, however you generate them). There will be
> slight geometry differences between SO3 and PO3, but a minimization
> should fix those pretty quickly

> if you read:
> not only "PO3" is 'missing': this is just a never ending problem ;-)

Yes of course... I just thougt that sulfate is there, so there also
would be phosphate.

> This is just the best choice and this was originally described by the
> Kollman's group in Cieplak et al. (but not for sugars):
> Sugar-OH Me-PO4(2-) -> Sugar + PO4(2-) -> Sugar-PO4(2-)
> <--->
> 1 inter-mcc
> Sugar-(OH)2 Me-PO4(2-) -> Sugar + PO4(2-) -> Sugar-[PO4(2-)]2
> 2 inter-mcc
> etc...
We will look into that.

>> However, as far as I understand things, we will need to
>> construct several different carbohydrate-phosphates and perform the ESP
>> fit process simultaneously for all of them, subject to the constraints that
>> -the partial charges of all atoms of the carbohydrate moiety retain
>> their original GLYCAM-defined value,
> 'retain GLYCAM-defined value'?
> Just create your own force field - why do you need to retain what others did?
> This is particularly true if you need to introduce different phosphate
> groups in your oligosaccharide: retaining GLYCAM-defined value will
> just introduce too many errors in the charge fitting step...
My train of thought was that the bond, torsion and dihedral terms used
in GLYCAM would by carefully balanced with the partial charges to
reproduce the certainly subtle details of the conformational equilibria
of the different carbohydrates. Therefore I assumed that changing those
would not be a good idea. I also assumed that in order to be used as a
"building block", all GLYCAM carbohydrate moieties would have the same
total charge, which obviously is not the case, as you mentioned in your
last email.
>> -the partial charges of all atoms of the phosphate moiety are the same
>> for each carbohydrate-phosphate and
> the charge of the monosaccharide units are different and the charges
> of the phosphate groups can be identical or different depending on the
> number of building blocks used for the phosphate groups; that being
> said - I would first simplify the problem and use a single building
> block for the phosphate groups...
Yes. But what I do not get is this: If each monosaccharide unit can have
a different total charge, we obviously can not use the same partial
charges for the phosphate group atoms when bound to different
monosaccharides and need to carry out RESP-fitting for all
monosaccharide-phosphates that occur in our simulation. For the sulfate
moiety as used in the carbohydrate builder, this is obviously not needed.

Thanks for all the information provided so far.

Jochen Heil,
TU Dortmund

AMBER mailing list
Received on Thu Jun 19 2014 - 05:00:02 PDT
Custom Search