On Thu, Mar 24, 2011 at 2:41 AM, Antony Oliver
Although I can see from the non-crystallogapher end-user perspective that the zero occupancy option is probably preferred, because as at least the side-chain atoms are preserved, and readily identifiable when working in PyMOL, CCP4MG, Coot etc.
But if the naive end user sees an arginine sidechain sticking out, he/she will most likely assume that the coordinates reflect reality. I'm inclined to agree with Frank - we're not making theoretical models here, so we shouldn't be guessing coordinate positions. Even in a 4A structure, however poor the sidechain density is, the coordinates have been refined against the data, and the B-factors will still provide some information about disorder and accuracy. If end users demand complete sidechains, let them make their own - this is far less confusing (and less dangerous) than making something up and expecting them to figure out what the zero column means. -Nat