Q: How to analyse water in an MD trajectory. I'm interested in being able to quantify the most frequently occupied water positions, and in particular the frequency of water mediated hydrogen bonds. A: The way I did it (which is discussed briefly in JACS 119, 4805-4825 (1997)) was really more to visualize where the water is rather than quantitate it or as a means to keep track of individual waters and their interactions (which is a bookkeeping nightmare). I first RMS fit the solute to a common reference frame, then construct a grid on this and count the number of particular atoms within each grid element (for example water oxygen atoms). This grid is dumped to the XPLOR density file format. The program midas contains a delegate (density) which can read this format. If you use midas, you can check to see if the delegate density is present by typing: delegate ddd start density If the console doesn't dump info about not being able to find "density" it should work... With this delegate, you can contour the grid to show iso-density levels. I typically display ~2-4x bulk water concentration. If you don't have midas with density, there is also skeleton code to dump out high probability grid elements into PDB format. The code can be changed easily enough to alter what probability this is (i.e. think now it dumps the highest 10% of grid elements). All of the grid/rms/imaging code is within the program "rdparm" and the newtransform routines. This is available with AMBER 5.0, although I could send a copy of the program (rdparm) to you if you don't have 5.0. IF you have 5.0, see the ptraj or rdparm notes and the HTML file there. [If you do use this program, you will have to be aware of some issues. The grid is slightly offset to the molecules reference frame by half the grid spacing. So, if you use midas, when you read in the pdb file corresponding to the grid then load up the grid, you will need to shift the pdb or the grid. I can explain this in more detail if you get this far... Also, if you do use the PDB dumping, I will have to refresh my memory to see if this is offset by 1/2 the grid spacing too. In some sense this all "works" but it isn't pretty or extremely user friendly.] Thomas Cheatham, III ---