[phenixbb] Bulk solvent correction

Nathaniel Echols nechols at lbl.gov
Thu Mar 31 08:05:17 PDT 2011


On Thu, Mar 31, 2011 at 7:27 AM, Ina Lindemann
<lindema2 at staff.uni-marburg.de> wrote:
> 1.Why is there such a big difference between R values with and without bulk
> solvent correction and anisotropic scale?
>
> Start R-work = 0.3765, R-free = 0.5498 (no bulk solvent and anisotropic
> scale)
> Final R-work = 0.3733, R-free = 0.5509 (no bulk solvent and anisotropic
> scale)
>
> Start R-work = 0.2356, R-free = 0.2962
> Final R-work = 0.2250, R-free = 0.2939

This is normal - the bulk solvent effect dominates at low resolution,
where the amplitudes are largest and thus contribute the most to the
R-factors.  If you were to turn off bulk solvent correction and
truncate your data at 8A or 6A, the gap will mostly disappear.  (Don't
actually do this, of course, but it appears to have been common
practice until around the mid-1990s.)

> 2. (the other structure) Why does the R value without bulk solvent and
> anisotropic scale decrease so much but with it remains stable?
>
> Start R-work = 0.2182, R-free = 0.2856 (no bulk solvent and anisotropic
> scale)
> Final R-work = 0.4054, R-free = 0.4223 (no bulk solvent and anisotropic
> scale)
>
> Start R-work = 0.2157, R-free = 0.2827
> Final R-work = 0.2147, R-free = 0.2816

Did you mean "increase", or is that first set of numbers backwards?
I'm not sure what the explanation is, but I'd advise against worrying
too much about the meaning of the unscaled R-factors.  (It may not be
a good idea for us to display them - it seems to create a lot of
confusion, especially since no other refinement program that I've used
does this.)

-Nat



More information about the phenixbb mailing list