Ariel Talavera wrote:
Hi all,
I am working with a 2.5 Å resolution structure with Wilson B factor
37,5 Ų. It is already 100 % built but after the TLS refinement the
average B factor was quiet low (4.6 Ų) and of course also the B
factors for each atom were also very low. For that reason I ran TLSANL
program to restore the total B factor. For each of the atoms the B
factors are now around 22-80 Ų but the average B (at the PDB file
header) was not updated (still 4.6). So, I reran Refmac, ( either TLS
& restrained refinement or restrained refinement with Fix TLS
parameters). After this the average B factor is now 24 Ų what seems
ok, but I did this blindly and I am afraid I am doing something wrong
by over doing the TLS refinement or whatever.
Any help will be more than appreciated.
Thanks a lot in advanced.
Best regards,
Ariel
Hello - I think we have met ?
Re B factors - you have found a bug in TLSANL - it should update the
average B but obviously doesnt.
There is a lot of discussion about how best to present TLS parameters,
and so far no proper agreement -
This isnt necessarily a fix, but my procedure is to start TLS again with
no parameters - just the group definitions and with the starting B
factor set to an average B - the default value is 20.
You would then expect the TLS B factors to more or less average out to 17.5
Then run TLSANL to get the complete b factor.
But TLSANL does not update the average B or correct the TLS parameters
in the header. the wwwPDB is debating how the information about TLS
should be presented - maybe you should query them?
Eleanor
