Perhaps I was confused by the "refinement exclude" keyword which explicitly says that atoms excluded from refinement will contribute to the mask calculation. Thanks for the correction. I would still object to the zero-occupancy atoms on semantic grounds. Partial occupancy means that an atom spends this fraction of time an this position. Zero occupancy thus says that the atom is never near this particular point in space, which for disordered side chains is likely untrue.
Cheers, Ed. > I agree that zero occupancy is a bit ugly, but useful when not sure > whether you will ever see that LYS.. > > But I dont think it wlll displace bulk solvent - at least not in REFMAC > where an atom with occ=0.0 will not contribute to the atom map. And I > expect this is true for all other structure factor calculations using > various programs.. > > Eleanor > > Ed Pozharski wrote: >> On Thu, 2010-08-12 at 08:57 +0000, MARTYN SYMMONS wrote: >>> Zero occupancy is generally a deprecated way of dealing with missing >>> density as it is confusing for less experienced user of the >>> coordinates. I think zero occupancy can be useful during refinement as >>> the atoms help fill space (or for example satisfy NCS restraint >>> format requirement) but then these atoms can be stripped out before >>> deposition. They should in any case never be included in B-factor >>> refinement as they will skew the statistics and possibly the B-factor >>> restraint model. >> >> Zero occupancy may be a bad idea for yet another reason - the atoms will >> displace bulk solvent and produce what is essentially a hole in the >> structure. It may be justified if you are trying to fill the empty >> internal cavities, but for atoms missing from density it seems like a >> wrong approach. >> > -- Edwin Pozharski, PhD University of Maryland, Baltimore
