Hi Daqi,

I refined my structure (of which the data is very anisotropic) using phenix.refine and with individual_sites, individual_adp, and tls options turned on. From the refined coordinates, the highest B-factor in the ANISO record (ie B_tls + B_individual) is 305. This number looks rather high. Is it something I should worry about? and is there an easy way to compute the individual B-factor component?

- what is "high", what is "low", what is "acceptably high",... ? This was discussed many times on both phenixbb and ccp4bb.

- you can use phenix.tls
For usage information just type phenix.tls and hit Enter, then read message output to the screen.
Using this command you can go back and forth between total and local B-factors, as defined here:

http://www.phenix-online.org/presentations/neutron_japan_2009/phenix_japan_part1.pdf
see slides #30-36.

Note 1) going from individual components Utls, Uatom to Utotal is the exact operation (numbers are computed exactly by the formula), while going back: splitting Utotal into Utls and Uatom is an approximate operation, since Utls is obtained as a least-squares fit of Utls to Utotal (Utls_fit), and the Uatom then is Utotal - Utls_fit.

Note 2) Uatom may be positive as well as negative or zero.

Please let me know if you have any questions or need help with this.

Pavel.