Position restraints can only be applied on a per-moleculetype basis, so the only atom numbers you can use are 1,2,3 to refer to O,H,H in the water. Your initial approach was pretty close - create a special moleculetype for the restrained waters and specify them as their own block in [molecules] in the topology. 

If you want to restrain the water oxygens to the x-y plane, your position restraint definition was simply wrong, as it restrains in all three spatial dimensions.  What you want is:

#ifdef POSRES_WATER_2

; Position restraint for each water oxygen

[ position_restraints ]

;  i funct       fcx        fcy        fcz

   1    1         0          0       1000

#endif

This allows waters to move freely in x and y (no restraint) but penalizes deviations along z.

Hi, 

I would extract the atom indices of the water molecules that you want to positionally restraint (crystal water if I understand you correctly). I would then create a new file called posre_crystal_hoh.itp that would look like this.

[ position_restraints ]

atom_index1 1 1000 1000 1000

atom_index2 1 1000 1000 1000

atom_index3 1 1000 1000 1000

........

and your topology would include an if statement calling this file like:

#ifdef POSRE_WATER_CRYSTAL

#include ./posre_crystal_hoh.itp ; with the file being in your working dir

#endif

In your mdp file then define:

define = -DPOSRE_WATER_CRYSTAL ; and anything else you would like to restrain

This should work? 

Hope this helps. 

Cheers,

Davide

Thanks for a detailed reply Davide. I followed it and made an index file of crystal water and then position restrain file. My index file starts from 1984 index (not from 1) to end. Now when i include the new .itp file (which again starts from 1984 index) it in .top  as 

I keep getting error as 

"Fatal error: [ file position_outside_water.itp, line 9 ]:

Atom index (1984) in position_restraints out of bounds (1-4).

This probably means that you have inserted topology section "position_restraints"in a part belonging to a different molecule than you intended to.In that case move the "position_restraints" section to the right molecule."

I searched online , tried different things but could not resolve it. The genrestr-h (from which i generated the new position restrain index file) reads a warning as :- "Since the atom numbers in every moleculetype in the topology start at 1 and the numbers in the input file for genrestr number consecutively from 1, genrestr will only produce a useful file for the first molecule."

Any help on how I should make the new position restrain for a molecule which is not FIRST

Not sure where you inserted the #include statement in the topology. It should be included after the molecule type that you want to positionally restrain: in this case water i guess. 

Otherwise, insert your #ifdef statement directly in the .itp file of the water molecules (tip4p2.itp). I would make a copy of the that file in your working directory and include it in the topology with #include "./tip4p2.itp".  

Hope this helps.

Cheers,

D.

Position restraints can only be applied on a per-moleculetype basis, so the only atom numbers you can use are 1,2,3 to refer to O,H,H in the water. Your initial approach was pretty close - create a special moleculetype for the restrained waters and specify them as their own block in [molecules] in the topology. 

If you want to restrain the water oxygens to the x-y plane, your position restraint definition was simply wrong, as it restrains in all three spatial dimensions.  What you want is:

#ifdef POSRES_WATER_2

; Position restraint for each water oxygen

[ position_restraints ]

;  i funct       fcx        fcy        fcz

   1    1         0          0       1000

#endif

This allows waters to move freely in x and y (no restraint) but penalizes deviations along z.

Thanks Justin for the response .Just for clarity, my  x & y   were not axis of restrain but as variables (or examples ) used to tell x and y number of water molecules :)

If I understood rightly , 

I want to restrain (lets say) water that start with index 50 to index 100 with each index number specifying a water atom like OW, H1 and H2 in my .gro file, then I can not use an index file with index starting from 50 to 100  but  rather just specify a  new section as follows 

#ifdef POSRES_WATER_2

; Position restraint for each water oxygen

[ position_restraints ]

;  i funct       fcx        fcy        fcz

   1    1         0          0       1000

#endif

and add an additional [ molecule ] type for my waters starting from 50 to 100 ?

Also another similar question (since I learn most of the things from your replies), how should I approach  to "Freeze" a sphere of radius r full of water in my MD run?  

Yes, any waters restrained using that [moleculetype] need to be defined as such in [molecules] and must be in the same order in the coordinate file.  Note, too, that you cannot use SETTLE for constraining the geometry in multiple blocks; this is a quirk of GROMACS, so in your SOL2 definition, you will need to edit the topology to have three manually defined [constraints] to replace the function of SETTLE.

Freezing a sphere becomes slightly more challenging, but really the same method applies - find the atoms that satisfy the geometric criteria using gmx select and do basically the same thing you're doing here.