[gmx-users] distorted phosphate group

[Christian Burisch]

Daniel Rigden wrote:
> Dear all

Hi Daniel,

> We're trying to do a simulation of a protein with bound
> 2-phosphoglycerate.  I've used PRODRG to get the itp file.  It does a
> great job, except for an unexpected problem with the phospho group - it
> becomes distorted after minimisation.

I had some problems with phosphate groups generated by PRODRG, too 
(the phosphates were distorted and sometimes even exploding during an 
MD). I always wanted to report this to Daan van Aalten, but 
unfortunately forgot it completely, now I hope he is reading this... ;-)

I solved this by using parts of the standard Gromos96 PDG description 
of protonated phosphates for my unprotonated ones (the charges are 
derived from DFT calculations; the PRODRG charges were IMHO too 
unpolar and did not match the standard groups of the force field, e.g. 
hydroxyl groups), and it seems to work fine for me (there are no 
impropers on the phosphate at all and the IMHO chemically equivalent 
oxygens are all type OM):

  [ atoms ]
  ; atom type   charge cgnr
    O6P  OM    -1.045  1
    O5P  OM    -1.045  1
    O4P  OM    -1.045  1
     P2   P     1.696  1
     O5  OA    -0.630  1
     C5 CH2     0.069  1
...

  [ bonds ]
  ;ai  aj   gromos
  O6P   P2  gb_23
   P2  O5P  gb_23
   P2  O4P  gb_23
   P2   O5  gb_27
   O5   C5  gb_17 ; maybe gb_19
...

  [ angles ]
  ;ai  aj  ak   gromos
   O6P P2  O5P  ga_28
   O6P P2  O4P  ga_28
   O5P P2  O4P  ga_28
   O6P P2  O5   ga_13
   O5P P2  O5   ga_13
   O4P P2  O5   ga_13
    P2 O5  C5   ga_25
...

  [ dihedrals ]
  ;ai  aj   ak   al  gromos
   C5  O5   P2  O6P  gd_11
   C5  O5   P2  O6P  gd_9
   P2  O5   C5   C4  gd_14
...

Regards

Christian