[phenixbb] WAS: changing TLS groups mid refinement

[Phil Jeffrey]

No.  You've missed the point here.


Pavel Afonine wrote:
> Dear Phil,
> 
> on a further thought about your "Longer version":
> 
>> Default behavior yields:
>> Final: r_work = 0.1898 r_free = 0.2479 bonds = 0.007 angles = 1.114
>> (...)
>> Truncate=yes data yields:
>> Final: r_work = 0.1932 r_free = 0.2473 bonds = 0.007 angles = 1.113
> 
> these Rfree-factor results above seem "identical" to me (we can compare 
> them since they were the same for both refinements).

Except they are absolutely not the same set of reflections. 
phenix.refine throws away the weakest data in the "default behavior" 
case.  It throws away 5.7% of them.  There are fewer reflections (read 
down the email and look at the Remark 3's from the PDB file).

Truncate includes this data in the truncate=yes case, which contains 
more reflections (see Remark 3), this additional data will be weaker, so 
you'd expect R-free and R-work to be higher.

I even agree that it's not wise to compare these two R-free's, which is 
why I created the data subset in the last test in my earlier email.

A more ideal test would be to remove the extra truncate=yes reflections 
from the test set.  However I don't know of any program that will do 
that (at least not with my current mtz file) and I'm not going to hand 
edit the data since I'm actually using this data.  Regenerating a test 
set would involve concerns about removing work set bias from the new 
test set, and that's a can of worms I do not want to enter into.

Phil