Questions and problems?

Edit

I want to use large NCUBE but can't - the program stops:
Requested water cube size exceeds MAXWAT = 14000

4.1 and later: change MAXWAT in sizes.h and recompile. (4.0: change resize.csh and recompile.) Further redimansioning/recompiling may be needed if other arrays are exceeded as a result.

Note that Leap does not have size restrictions.

I used NCUBE=2 in AMBER4.0. But in 4.1, it said that I have to use NCUBE=3 for the same molecule.

The old Edit allowed you to set NCUBE and CUTX,CUTY,CUTZ incompatibly so that your box would not be filled with water out to the specified edges. It is fixed in 4.1 so that Edit tells you when it doesn't have enough waters to fill the box you define. If you want to use fewer waters, set your CUTX,CUTY,CUTZ parameters smaller.

Unfilled boxes due to too small NCUBE generated by 4.0 Edit would shrink in constant pressure dynamics. In constant volume dynamics, there would probably be bubble-like VDW voids. If this is desired, the box could be manually expanded by editing prmtop (box coords are at the end).

I use NCUBE=4 in the EDIT module. It creates a box size of 75 A. This I feel is too big for my molecule.

NCUBE=4 is not what is causing the large size: you must set CUTX, CUTY, CUTZ. As the manual says:

         CUTX       Any water molecule which is farther than "CUTX" along
              the X-axis from any atom of solute will be discarded
              for the Calculation. Essentially CUTX defines the
              minimum thickness of the solvent shell along the X-axis.

Forcing the user to specify NCUBE was a bad design decision made many years ago. The program should determine on its own how many boxes are required in each dimension to satisfy the user-requested solvation clearances. This is what leap does - in fact the whole link/edit/parm sequence in leap looks like this:

% tleap
> x = loadpdb my_molecule.pdb
> solvatebox x WATBOX216 10
> saveamberparm x my_mol.top my_mol.crd
> quit

Also after running the EDIT can I get an Pdb output for the molecule with the box of water...so that I can view it on a graphics package.

In xleap, the same again, allowing you to watch the waters being added as well as saving the pdb:

% xleap
> x = loadpdb my_molecule.pdb
> edit x
> solvatebox x WATBOX216 10
> savepdb x my_mol_solv.pdb
> saveamberparm x my_mol.top my_mol.crd
> quit

Also, EDIT can generate pdb files directly, and pdbgen or ambpdb can be used to generate pdb from prmtop+inpcrd/restrt.

The box size of the original wat216.dat box is 18.77A, but after running Edit, the output file states that the box size is now 19.17A. I.e. a difference of 0.4A is added.

Is there any reason for adding 0.4A to the box size?

From the Release notes:


    Edit:

        When  adding  a  box  trimmed to a certain distance from a
    molecule, 0.4 A clearance is now added at the box  boundary  to
    prevent  close  water-water contacts, since Edit does not check
    for them. This decreases the possibility of  shake  failure  at
    minimization or dynamics startup.

When generating an arbitrary box bigger than the template wat216-box, the density is way to low: If you generate a box of 25A*25A*25A with the WAT option, 466 water molecules are included in the box - so the density is 0.891 g/cm3. The standard water box (wat216.dat) has a density of 0.976 g/cm3. So the density decreases approx. 9%.

Without doubt, the density plays an important role in constant volume simulations of solvated systems, so this is something I definitely want to get correct.

As for the problem above, there is no way to have correct vdw contacts at the box boundary when using a non-integral number of boxes. (This is why leap has the option of solvating with integral boxes, although in my opinion it is not worth the effort.) Solvating a molecule also further decreases the density, because of vdw voids where partially touching waters are removed.

The only way to get correct density is by doing your own equilibration using constant pressure. Then you can switch to constant volume if you want.

No matter how big I make MAXWAT in sizes.h and increase NCUBE in edit.in, EDIT seems to want too many boxes:

 ** SOLUTE + X CLEARANCE = 301.625: EXCEEDS TOTAL BOX DIMENSION:  75.097

If EDIT is forced to generate coordinates for the first atoms in a chain (usually hydrogen), it uses internal coordinates as cartesian, e.g. x=1, y=120, z=0. This typically creates one or more atoms distantly removed from the solute.

If you take a look at the PDB file generated from EDIT without water (always a good idea to monitor it and take a close look), you will see this. The "standard" way to fix this is by "ABC" option in EDIT. However, some found it difficult to use and it generally involves a few spatial operations (like translation and rotation) in the head. The easy ways to fix the problem are,

If the problem is with the hydrogen atoms, you can use Leap instead of EDIT, or try PROTONATE (should work for standard residues).
If the problem is with the first few heavy atoms (e.g., the first residue is absent in the X-ray PDB, happens when you add ACE at the begining), you need to fix them manually, unfortunately. (This applies to EDIT, and possibly Leap if the whole first residue is missing.) Alternatively in such cases, one could allow the program to place the atoms using the "bogus" coordinates, then energy minimize in vacuum before adding waters, keeping the other atoms fixed by using the belly option on the added atoms. In this case, use of SHAKE on all bonds is recommended when minimizing, since it can pull the atoms into place "through" other atoms if necessary (xleap's minimizer also can be used, since it does not take nonbonded interactions into account). Caution should be taken when using this approach to add chiral atoms, since residue template information may not be sufficient to make the right choice: in this case, manual operation is strongly recommended.
If the problem is with counterions (this only happens with EDIT when the ions are defined in LINK), you can add them manually to the pdb, or try the cion program.

Yong Duan and Bill Ross