Hi Mark,
1. If I am refining TLS and individual isotropic ADPs, does phenix.refine derive iso-ADP from the ANISOU records presently in the file, or does it use the iso-ADP from the ATOM record and recombine it with the TLS values?
phenix.refine always outputs total ADP, both isotropic in ATOM and anisotropic in ANISOU records.
2. Consequently if I reset the B's for a range of residues on the ATOM cards only, will they be used rather than the ANISOU record to derive the ADP?
ANISOU are used if present. So if you want to reset B-factors, then first you need to convert to isotropic (or just discard ANISOUs) and then set isotropic B to a desired value.
3. Should phenix.refine input files retain the header lines containing the previously cycle TLS parameters?
phenix.refine does not use values of T, L and S matrices from PDB file header.
4. If one explicitly changes atom ADPs with Coot using Measures>Residue info... the output file reflects the change only in the ATOM lines, but the ANISOU line retains the previous 6 parameters. It seems as if coot should also change the ANISOU to reflect anisou (last 3 parameters=0).
Indeed it's a good idea to change ANISOU as well to be consistent. Pavel