The problem: Extremely low Bfactor values for water molecules after TLS refinement (Bfact. = 2) despite being associated with sidechains with much higher Bfactors. The problem apparently arises because REFMAC includes the waters in TLS groups by default.
The solution: exlude the waters from TLS refinement using the command: TLSD waters exclude (thanks Roberto Steiner). observations of a ludite: I was adding the TLSD command using the "run&view com file", in what i thought was the logical place (after the REFI-, and ncyc commands), but interestingly REFMAC appeared to forget about my weight- MATRIX value and the bond/angles went a bit crazy. when added after the weight command things look a bit more normal for the bonds/angles, as well as the water Bfactors. don't know why this should be- maybe it is just Monday messing with me........ Hope this helps, Al
