[gmx-users] Atom types for phosphate group in OPLS-AA force field

[Justin A. Lemkul]

On 4/26/12 7:45 AM, Delmotte, Antoine wrote:
> Dear Gromacs users,
>
>
> I am here requesting your help with regards to the editing of the OPLS-AA force
> field, and more specifically, the choice of the atom types.
>
> I have added an entry for the molecule CABP to the .rtp and .hdb files.
>
> I can run pdb2gmx, genbox and editconf without getting any error. However, when
> I run grompp, I receive the following errors:
>
> ERROR 1 [file 1RBO_Protein_chain_A.itp, line 15054]:
> No default Bond types
>
> ERROR 78 [file 1RBO_Protein_chain_M.itp, line 47276]:
> No default Angle types
>
> ERROR 82 [file 1RBO_Protein_chain_M.itp, line 66639]:
> No default Ryckaert-Bell. types
>
> (there are 160 of those).
>
>
>  From what I could see in this mailing list, it seems the error comes from the
> choice of the atom types for the phosphate groups. I have tried several of them
> which seemed to make sense, but none of them allowed to remove all the errors.
>
> The end of the chain is something like:
>
> O
> |
> (...)--CH2--O--P--O--H
> |
> O
>
> (details of the whole molecule here:
> http://www.rcsb.org/pdb/ligand/ligandsummary.do?hetId=CAP&sid=1RBO
> <http://www.rcsb.org/pdb/ligand/ligandsummary.do?hetId=CAP&sid=1RBO> )
>
> The problem is about the atom type of the -OH group at the end. Here are the
> entries from the .atp file which I thought were relevant:
>
> opls_445 30.97376 ; P in MeOPO3--, MeOPO3H2
> opls_446 15.99940 ; O= in MeOPO3--, MeOPO3H2
> opls_447 15.99940 ; OMe in MeOPO3--, MeOPO3H2 methyl phosphate
> opls_448 12.01100 ; C in MeOPO3--, MeOPO3H2 6-31+G* CHELPG
> opls_449 1.00800 ; H in MeOPO3--, MeOPO3H2
>
> I chose opls_445 for the P, opls_446 for the 3 oxygens, and opls_449 for the
> hydrogen. Apparently, I got it wrong for the oxygen and the hydrogen of the OH
> group, at least. Unless the error actually comes from something else...
>
> So my question is the following: Have I done something wrong about the atom
> types? If so, does anyone know which atom type I should choose?
>

The problem isn't necessarily the atom types themselves (which seem to be 
reasonable enough from the information provided), but rather that there are no 
default bonded parameters for these interactions.  If you look into 
ffnonbonded.itp, the second column is entitled "bond_type" and thus refers to 
the names used by grompp when interpreting how atoms are connected.  The actual 
bond, angle, and dihedral parameters are in ffbonded.itp, using these names. 
You would have to map out what the bond_type names are for each of these and 
find/derive new parameters for each of the interactions that are missing.

-Justin

> Details of my rtp entry:
>
> [ CAP ]
> [ atoms ]
> P1 opls_445 1.539429 1
> O1P opls_446 −0.818751 1
> O2P opls_446 −0.899293 1
> HO2P opls_449 0.429028 1
> O3P opls_446 −0.895147 1
> O1 opls_447 −0.772438 2
> C1 opls_448 0.022084 3
> HC11 opls_449 0.176920 3
> HC12 opls_449 0.139342 3
> C2 opls_159 0.146383 4
> O2 opls_154 −0.883774 4
> HO2 opls_155 0.527657 4
> C opls_271 0.815849 5
> O6 opls_271 −0.766517 5
> O7 opls_272 −0.771422 5
> C3 opls_158 0.142583 6
> HC3 opls_140 0.107158 6
> O3 opls_154 −0.817159 6
> HO3 opls_155 0.471262 6
> C4 opls_158 0.184473 7
> HC4 opls_140 0.164192 7
> O4 opls_154 −0.822792 7
> HO4 opls_155 0.498509 7
> C5 opls_448 0.038563 8
> HC51 opls_449 0.143027 8
> HC52 opls_449 0.135804 8
> O5 opls_447 −0.730025 9
> P2 opls_445 1.526672 10
> O4P opls_446 −0.866625 10
> O5P opls_446 −0.810148 10
> HO5P opls_449 0.521464 10
> O6P opls_446 −0.876311 10
> [ bonds ]
> HO2P O2P
> O1P P1
> O2P P2
> O3P P1
> P1 O1
> O1 C1
> C1 HC11
> C1 HC12
> C1 C2
> C2 O2
> O2 HO2
> C2 C
> C O6
> C O7
> C2 C3
> C3 O3
> C3 HC3
> O3 HO3
> C3 C4
> C4 HC4
> C4 O4
> O4 HO4
> C4 C5
> C5 HC51
> C5 HC52
> C5 O5
> O5 P2
> P2 O4P
> P2 O5P
> P2 O6P
> O5P HO5P
>
> Thank you very much in advance!
>
> Antoine
>
>

-- 
========================================

Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
MILES-IGERT Trainee
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================