Hi Hailiang, phenix.model_vs_data handles input TLS exactly as described in the paper you mentioned below - I didn't change anything in the relevant code since publication of that paper. Essentially, there may be three types of input files: - containing total ADP, and - containing residual (or local or atom) ADP, - various weird mixtures of two above, so without trusting too much input REMARK 3 records, phenix.model_vs_data tries both possibilities and picks the one that yields a lower R-factor, which makes the program a bit slower but way more robust. Pavel.
In phenix.model_vs_data, a special handling of ANISOU records and TLS parameters were mentioned in "http://scripts.iucr.org/cgi-bin/paper?he5476". My understanding is the following and pls let me know if I am wrong:
************ (1) In TLS, some programs output a PDB with ANISOU including only TLS B-factor (case A), and other programs output a PDB with ANISOU including both TLS and residual B-factor (case B).
(2) In the former case (A), the ANISOU B will be added to the residual B; The later case (B) is straightforward. (Both will be tried and lowest R/Rfree will be used)
(3) In the former case (A), if ATOM B records contain both TLS (isotropic conversion) and residual B, TLS-isotropic B will be substracted from ATOM B records. The later case (B) is straightforward. (Both will be tried and lowest R/Rfree will be used)
*************
Pls let me know if I am wrong somewhere. Thanks a lot!
Best Regards, Hailiang
_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb