On Tue, Apr 3, 2012 at 8:50 AM, Pavel Afonine
are these waters agree with the density? All phenix.refine is using to decide about whether to keep or remove a water is: - density (mFo-DFc and 2mFo-DFc) peak height and shape; - distance between the density peak and surrounding atoms. The corresponding control parameters are exposed to users, so one can make them softer or stricter. If phenix.refine deletes a water that only means that one or a few selection criteria are not fulfilled.
It is often not obvious which of these criteria need to be changed. Using the default settings, I always see additional peaks left over that to my eye look like waters, and it almost always removes waters that I am convinced were placed correctly. I agree with John - it would be nice to be able to specify that the routine should only add waters and not remove any that were in the input file. At the very least, it should not be removing waters that have bond restraints. -Nat