modified on 29 November 2011 at 18:31 ••• 22,546 views

QMMM Appendix

From NWChem

Jump to: navigation, search

Contents

Preparing QM/MM calculations from scratch

Setting up QM/MM calculations for a new system for which classical force analog is not readily available would typically involve the following steps

  1. Generation of the proper PDB file
  2. Construction the fragment file(s)
  3. Creation of parameter file if new atom types were defined

Generation of the proper PDB file

It is often the case that the input structure for the system comes in the form of the xyz file. Let us take a concrete example of N3O3- molecule, which we would like to embed in classical solvent and perform QM/MM calculations. Here is the structure of just N3O3- in xyz format (generated in course of gas phase optimizations)

   6    
  geometry 
  N                     1.31562667     0.93574165    -0.42424728
  O                     1.56161766     0.18015298    -1.36827899
  N                     2.36373381     1.02559495     0.48834956
  O                     3.47240000     0.42852552     0.42137570
  N                     1.95804013     1.90608355     1.48799418
  O                     2.81393172     2.06788134     2.36142683

We cannot use this file as is in the QM/MM simulations, and it has to be converted into PDB format. This is needed even if we plan to treat this molecule quantum mechanically. There is more than way to do it. For example, we could use Babel (http://openbabel.org/wiki/Main_Page), which will generate the PDB file as

COMPND    geometry 
AUTHOR    GENERATED BY OPEN BABEL 2.3.0
HETATM    1  O   LIG     1       1.562   0.180  -1.368  1.00  0.00           O  
HETATM    2  N   LIG     1       1.316   0.936  -0.424  1.00  0.00           N  
HETATM    3  N   LIG     1       2.364   1.026   0.488  1.00  0.00           N  
HETATM    4  O   LIG     1       3.472   0.429   0.421  1.00  0.00           O  
HETATM    5  N   LIG     1       1.958   1.906   1.488  1.00  0.00           N  
HETATM    6  O   LIG     1       2.814   2.068   2.361  1.00  0.00           O  
CONECT    1    2                                                      
CONECT    2    1    3                                                 
CONECT    3    2    4    5                                            
CONECT    4    3                                                      
CONECT    5    3    6                                                 
CONECT    6    5                                                      
MASTER        0    0    0    0    0    0    0    0    6    0    6    0
END

which after stripping nonessential information becomes

 HETATM    1  O   LIG     1       1.562   0.180  -1.368  1.00  0.00           O  
 HETATM    2  N   LIG     1       1.316   0.936  -0.424  1.00  0.00           N  
 HETATM    3  N   LIG     1       2.364   1.026   0.488  1.00  0.00           N  
 HETATM    4  O   LIG     1       3.472   0.429   0.421  1.00  0.00           O  
 HETATM    5  N   LIG     1       1.958   1.906   1.488  1.00  0.00           N  
 HETATM    6  O   LIG     1       2.814   2.068   2.361  1.00  0.00           O  
 END

This is not yet the format we want. So what we need to do is

1.replace HETATM field by ATOM preserving the format (column locations) of the rest of the file:

I would typically use sed for this purpose

 sed 's/HETATM/ATOM  /' n3o3-bad.pdb > n3o3-step1.pdb

where n3o3-bad.pdb is the original pdb file from Babel and n3o3-step1.pdb is the converted one as shown below

ATOM      1  O   LIG     1       1.562   0.180  -1.368  1.00  0.00           O
ATOM      2  N   LIG     1       1.316   0.936  -0.424  1.00  0.00           N
ATOM      3  N   LIG     1       2.364   1.026   0.488  1.00  0.00           N
ATOM      4  O   LIG     1       3.472   0.429   0.421  1.00  0.00           O
ATOM      5  N   LIG     1       1.958   1.906   1.488  1.00  0.00           N
ATOM      6  O   LIG     1       2.814   2.068   2.361  1.00  0.00           O
END

2. define residues (aka molecules):

In our case Babel did this for us and the entire system is defined as one residue with the name LIG (see columns 4 and 5). We can leave it as, but I will redefine residue name to NN3 (keep it to 3 character!). Again running sed 's/LIG/NN3/' n3o3-step1.pdb > n3o3-step2.pdb

ATOM      1  O   NN3     1       1.562   0.180  -1.368  1.00  0.00           O
ATOM      2  N   NN3     1       1.316   0.936  -0.424  1.00  0.00           N
ATOM      3  N   NN3     1       2.364   1.026   0.488  1.00  0.00           N
ATOM      4  O   NN3     1       3.472   0.429   0.421  1.00  0.00           O
ATOM      5  N   NN3     1       1.958   1.906   1.488  1.00  0.00           N
ATOM      6  O   NN3     1       2.814   2.068   2.361  1.00  0.00           O
END

You could have also broken it up into several residues as

ATOM      1  O   NN2     1       1.562   0.180  -1.368  1.00  0.00           O
ATOM      2  N   NN2     1       1.316   0.936  -0.424  1.00  0.00           N
ATOM      3  N   NN2     1       2.364   1.026   0.488  1.00  0.00           N
ATOM      4  O   NN2     1       3.472   0.429   0.421  1.00  0.00           O
ATOM      5  N   NN1     2       1.958   1.906   1.488  1.00  0.00           N
ATOM      6  O   NN1     2       2.814   2.068   2.361  1.00  0.00           O
END

where I defined two residues NN2 and NN1 (note changes in columns 4 and 5). To keep things simple I will stay with one residue version for now

3. make unique atom names:

This has to do with the requirement that all atoms names have to be unique within a given residue. So if we take our one residue version it could be modified as (notice column 3)

ATOM      1  O1  NN3     1       1.562   0.180  -1.368  1.00  0.00           O
ATOM      2  N1  NN3     1       1.316   0.936  -0.424  1.00  0.00           N
ATOM      3  N2  NN3     1       2.364   1.026   0.488  1.00  0.00           N
ATOM      4  O2  NN3     1       3.472   0.429   0.421  1.00  0.00           O
ATOM      5  N3  NN3     2       1.958   1.906   1.488  1.00  0.00           N
ATOM      6  O3  NN3     2       2.814   2.068   2.361  1.00  0.00           O
END


Now we have a PDB file for our system in "proper" PDB format. Before we move to next step, I should mention that the last 3 columns are not necessary and could have been removed at any point leading to n3o3.pdb

ATOM      1  O1  NN3     1       1.562   0.180  -1.368    
ATOM      2  N1  NN3     1       1.316   0.936  -0.424    
ATOM      3  N2  NN3     1       2.364   1.026   0.488    
ATOM      4  O2  NN3     1       3.472   0.429   0.421    
ATOM      5  N3  NN3     2       1.958   1.906   1.488    
ATOM      6  O3  NN3     2       2.814   2.068   2.361    
END

Generation of new fragment files

While setting up QM/MM calculations is often necessary to generate new fragment files for the molecules that are not available as part of standard set. The IMPORTANT assumption here is that these new molecules/residues will be part of OQM region, and as a result only minimum information needs to be provided to include them in QM/MM calculations.

First we must ensure that we have a proper PDB format for our as discussed in the Generation of the proper PDB file section. As a concrete example we will start with N3O3 example that was discussed there as well

ATOM      1  O1  NN3     1       1.562   0.180  -1.368    
ATOM      2  N1  NN3     1       1.316   0.936  -0.424    
ATOM      3  N2  NN3     1       2.364   1.026   0.488    
ATOM      4  O2  NN3     1       3.472   0.429   0.421    
ATOM      5  N3  NN3     1       1.958   1.906   1.488    
ATOM      6  O3  NN3     1       2.814   2.068   2.361    
END

We have residue named NN3 and therefore looking to construct fragment file NN3.frg. The best way to do it is to run a simple prepare job

start n3o3
prepare
source n3o3.pdb
new_top new_seq
new_rst
modify segment 1 quantum 
update lists
ignore
write n3o3_ref.pdb
write n3o3_ref.rst
end
task prepare

This prepare job will necessarily fail because NN3.frg is not yet available:

Created fragment                      ./NN3.frg_TMP
Unresolved atom types in fragment NN3
 **********
 *   0: pre_mkfrg failed 9999
 **********


As part of this process skeleton fragment file (NN3.frg_TMP) will be generated that can be modified into the final correct form. Let us take a look at NN3.frg_TMP


# This is an automatically generated fragment file
# Atom types and connectivity were derived from coordinates
# Atomic partial charges are crude estimates
# 00/00/00  00:00:00  
#
$NN3
    6    1    1    0
NN3
    1 O1             0    0    0    1    1    0.000000    0.000000
    2 N1             0    0    0    1    1    0.000000    0.000000
    3 N2             0    0    0    1    1    0.000000    0.000000
    4 O2             0    0    0    1    1    0.000000    0.000000
    5 N3             0    0    0    1    1    0.000000    0.000000
    6 O3             0    0    0    1    1    0.000000    0.000000
    1    2
    2    3
    3    4
    3    5
    5    6


The main problem with this fragment file is that there are no atom types to be found in column 12. What atom type does is to identify what classical parameters should be assigned to it. Since, as mentioned in the beginning, we are planing to treat this residue/molecule as part of QM region, the only classical information needed is VDW parameters. We will assume that all nitrogens atoms in our molecule can use the same set of parameters, and the same for oxygens. Therefore we will define two atom types NX and OX

$NN3
    6    1    1    0
NN3
    1 O1   OX        0    0    0    1    1    0.000000    0.000000
    2 N1   NX        0    0    0    1    1    0.000000    0.000000
    3 N2   NX        0    0    0    1    1    0.000000    0.000000
    4 O2   OX        0    0    0    1    1    0.000000    0.000000
    5 N3   NX        0    0    0    1    1    0.000000    0.000000
    6 O3   OX        0    0    0    1    1    0.000000    0.000000
    1    2
    2    3
    3    4
    3    5
    5    6

and rename resulting file as NN3.frg. Please note that the overall format of the fragment file was preserved and atom types were entered starting at column 12. Rerunning the same prepare job moves as a bit further this time

 modify segment     1 set 0 quantum
 Parameter file                        /Users/marat/opt/codes/nwchem-new/src/data/amber_s/amber.par
 Parameter file                        /Users/marat/opt/codes/nwchem-new/src/data/amber_q/amber.par
 Parameter file                        /Users/marat/opt/codes/nwchem-new/src/data/amber_x/amber.par
 Parameter file                        /Users/marat/opt/codes/nwchem-new/src/data/amber_u/amber.par


 Undetermined force field parameters
 Parameters could not be found for atom type     OX   Q
 Parameters could not be found for atom type     NX   Q
 **********
 *   0: pre_check failed 9999
 **********

complaining now that atom types OX and NX are not defined. This brings us to the next step of defining new parameter file for our calculation.

Generation of new parameter files

Continuing with our fragment construction in Generation of new fragment files section, we now need to define VDW parameters for our new atom types NX and OX. The best way to do it is to create amber.par file in the directory where you plan to rerun final prepare

Electrostatic 1-4 scaling factor     0.833333
Relative dielectric constant     1.000000
Parameters epsilon R*
#
Atoms
NX     14.01000 7.11280E-01 1.82400E-01                            1 1111111111
      Q       7 3.55640E-01 1.82400E-01
OX     16.00000 6.35968E-01 1.76830E-01                            1 1111111111
      Q       8 3.17984E-01 1.76830E-01
End

The format of this file is documented in Format of NWChem parameter file. How to actually choose the appropriate values for VDW parameters is a whole new subject, which I do not think anybody yet fully addressed. The practical strategy is to copy from known atom types, which are chemically similar to the ones in your system. In the case above I copied parameters from standard AMBER atom types N and OW.

Format of NWChem parameter file

The format of NWChem parameter is illustrated on the figure below and also available as pdf file.

Format of NWChem parameter file


Conversion of standard AMBER program parameter files

Fortran code that performs conversion from AMBER program parameter file format to NWChem can be found here. It works by parsing out free format AMBER style parameter file contained in amber.in

MASS
C  12.01
CA    12.01
BOND
#this is a comment
C -CA  469.0    1.409           this is also a comment
C  -   CB     447.0      1.419  
ANGLE
C -CA-CA    63.0      120.00 another comment
C -CB-NB    70.0      130.00
DIHEDRAL
X -C -CA-X    4   14.50        180.0             2.         intrpol.bsd.on C6H6
 X - C -  CB -X    4   12.00        180.0             2.         intrpol.bsd.on C6H6
IMPROPER
X -CT-N -CT         1.0          180.          2.           JCC,7,(1986),230
CT -O  - C -OH         10.5         180.           2.
NONBOND
 CA 1.9080  0.0860
C 1.9080  0.0860

to fixed format NWChem style amber.par file

#Generated amber.par file
Electrostatic 1-4 scaling factor     0.833333
Relative dielectric constant     1.000000
Parameters epsilon R*
#
Atoms
CA     12.01000 3.59824E-01 1.90800E-01                            1 1111111111
              6 1.79912E-01 1.90800E-01
C      12.01000 3.59824E-01 1.90800E-01                            1 1111111111
              6 1.79912E-01 1.90800E-01
Bonds
C    -CA      0.14090 3.92459E+05
C    -CB      0.14190 3.74050E+05
Angles
C    -CA   -CA      2.09440 5.27184E+02
C    -CB   -NB      2.26893 5.85760E+02
Proper dihedrals
     -C    -CA   -        3.14159 1.51670E+01    2
     -C    -CB   -        3.14159 1.25520E+01    2
Improper dihedrals
     -CT   -N    -CT      3.14159 4.18400E+00    2
CT   -O    -C    -OH      3.14159 4.39320E+01    2
End