I have a protein crystal (space group P6_5) with large water channels which I have set up using the CRYST commands. I believe I have eliminated all problems to do with coordinate alignment etc. (output PDB file with images look fine), and now wish to do some dynamics.
The problem is that during dynamics, water molecules leave the central asymmetric unit and the simulation crashes. In the past, I have always used various IMAGE centering commands (without the CRYST commands), but I was of the understanding that the CRYST commands should take care of this. However it seems to be BYSEG rather than BYRES, which is fine for small crystals (like the Ala example files), but not when many (mobile) water molecules are present.
Using additional IMAGE commands (IMAGE BYRES XCEN .. YCEN .. ZCEN .. SELE ( SEGID WAT ) END) doesn't seem to solve the problem. I'm just doing NVE dynamics at the moment, and am using 'IMGFRQ 10 INBFRQ -1'.
Any help, suggestions or comments would be very useful! David.
CRYSTal provides a convenient way to set up complicated image transformations, but the user must still specify how image centering should occur. The default is to center around 0,0,0 which may be a problem in your case. It may also be a problem if the asymmetric unit was moved to be at the center.
IMAGE BYRES should work fine for water, where each molecule is a residue, given the appropriate center.
Still having some problems with my hexagonal unit cell... I've now sorted out my image centering and a few other small errors in my setup, but there's still an odd feature which is worrying me.
The hexagonal unit cell has sides of length A and C. My water box (asymmetric unit) is a diamond shape, long diagonal A, short diagonal A/sqrt(3) and height C/2. I've got the appropriate number of water molecules in the box for the correct density. On minimisation and equilibration (NVE at the moment), the sides of the box bulge out in a symmetric way, approximately 1/3 up the height on two opposite sides, and 2/3 up the box height on the other two sides. The top and bottom of the box are flat and parallel, as they should be.
The positions of the bulges did make me think that something might be wrong with the C spacing, but I've checked this. Also, printing out a pdb with images shows no overlapping parts, nor gaps along A or C. This 'additional volume' is also a problem, since water is moving away from the sharp point of the asymmetric unit, meaning there is a 'hole' in the water.
I could perhaps solve this by generating the whole unit cell, but I want to avoid having to simulate 24 proteins and all the water if I can.
Here are some pieces of input to do with the crystal, which work fine when applied to the protein part of the crystal -- ie the lattice looks fine: ---- SET A 128.24 SET C 68.32 SET ALPHA 90 SET GAMMA 120