[gmx-users] Using dummies in lipids

[Senthil Kandasamy]

Maybe Anton can help me with this one.

I want to use a dummy construction for POPE so that I can use a larger
time step. Using united atoms for aliphatic hydrogens, the only explict
hydrogens in the POPE are in the headgroup. It is basically an NH3- . So
I used a dummy construction identical to that of a lysine.

The original itp file (from Tieleman) looks like 

[ atoms ]
;   nr    type   resnr  residu    atom    cgnr        charge         
mass
     1     H         1    POPE      H1       0         0.4000   1.0080 
     2     H         1    POPE      H2       0         0.4000   1.0080 
     3     H         1    POPE      H3       0         0.4000   1.0008 
     4     LNL       1    POPE      N4       0        -0.5000   14.0067
     5     LH2       1    POPE      C5       0         0.3000   14.0270 
     6     LC2       1    POPE      C6       1         0.4000   14.0270 
		etc...

The total number of atoms in the undummified topology is 52.

So, I added two dummied as atom numbers 53 and 54 to create  a modified
itp file.

_______________________________________________________________________________________
[ atoms ]
;   nr    type   resnr  residu    atom    cgnr        charge         
mass
     1     H         1    POE       H1       0         0.4000   0.00000 
     2     H         1    POE       H2       0         0.4000   0.00000 
     3     H         1    POE       H3       0         0.4000   0.00000 
     4     LNL       1    POE       N4       0        -0.5000   0.00000 
     5     LH2       1    POE       C5       0         0.3000   14.0270
...
    53	   MNH3	     1	  POE	   MZ1	    36             0    8.51535;
    54     MNH3	     1    POE      MZ2      37             0    8.51535;

...........other directives omitted.......................


 [constraints]

	5	53	2
	5	54	2
	53	54	2

[dummies3] 
4	5	53	54	1
1	5	53	54	1

[dummies3]

2	5	53	54	-4
3	5	53	54	4
________________________________________


I basically took the original topolgy, added the two MNH3 atoms and
added the constraints and dummies3 sections.

Then, I tried to grompp on a single dummified popc molecule. I created
approximate coordinates for MZ1 and MZ2 assuming that grompp will fix it
because of the constraints.

I got the following error.

WARNING 1 [file "pope_dum.itp", line 118]:
  No default Constr. No Conn. types, using zeroes
WARNING 2 [file "pope_dum.itp", line 119]:
  No default Constr. No Conn. types, using zeroes
Excluding 3 bonded neighbours for POE 1
processing coordinates...
double-checking input for internal consistency...
Calculating parameters for dummy atoms
Fatal error: Automatic parameter generation not supported for Dummy3
atom 4 for this bonding configuration


This is probably because atom 4 is of type LH2 which is not defined in
the constrainttype section of ffgmxbon.itp. Once I added a line, 

 MNH3  LH2    2   DC_MNC2


in the [constrainttype] section, assuming LH2 is similar to CH2 , the
warnings disappeared. However, the fatal error still remains :
_______________________________________________________
processing coordinates...
double-checking input for internal consistency...
Calculating parameters for dummy atoms
Fatal error: Automatic parameter generation not supported for Dummy3
atom 4 for this bonding configuration
[224] [3:39pm] [seveniron] ~/work/create/dummies/pope --->
__________________________________________________________


Am I doing something wrong here? Is there more that I need to do?
Hope someone can help me with this. 


Thanks.

-- 
Senthil Kandasamy <senthilk at umich.edu>