[phenixbb] coot links on Mac -> Possible MR solution

Vellieux Frederic Frederic.Vellieux at ibs.fr
Mon Aug 2 01:57:54 PDT 2010

vandana wrote:
> hello
> i have diffraction of a crystal showing Rmerge 50 % resolution 21- 3.5 
> A. this is due to crystal decay with time while data collection . 
> multiplicilty is 11 .0  and data is 99.6 % complete . i want to know 
> that is this a good data to solve structure with MR . i got solution 
> with this data but the R factor is 46.5 and R free is 51.4 .
> i want to know that if R merge is too high  we should proceed to solve 
> structure or not .

 From "X-ray structure determination, a practical guide" (Stout, G.H. & 
Jensen, L.H. citing Wilson A.J.C. [1950], Acta Cryst. 3, 397), the 
expected R-factors for the atoms of the structure randomly distributed 
in the unit cell (or asymmetric unit) are 0.83 for centric reflections 
and 0.59 for acentric reflections.

The values you indicate are well below the figures mentioned above, so 
they indicate that what you have is not random. But this is not 
sufficient to tell you if what you have is a proper MR solution or not:

What about the packing in the unit cell? Do you have crystal-forming 
contacts in the 3 directions of space that explain the formation of the 
crystal? If there are voids, there might be molecule(s) missing. What is 
the value of the Z-score provided by Phaser (if you have used Phaser)? 
In the resulting electron density map, do you see electron density that 
correspond to the "mutations" you have to carry out in order to go from 
model to target structure? If you see such electron density, then it is 
very likely that you indeed have a molecular replacement solution.

The value of the Rmerge (you probably mean Rsym) you give (0.5) will 
probably not satisfy referees for a publication. What can always be done 
is to use several crystals for data collection, starting data collection 
at different (unique) positions, and merge the resulting data 
(restricting the data processing to the frames where the value of Rsym 
is acceptable). You should then end up with a merged data set that has a 
lower Rmerge value. We used to do this all the time in the days prior to 
cryo-cooling of crystals, when crystals were mounted in capillaries. 
Rmerge values of 10 to 15 % were typical then. That is if the value of 
the Rsym is indeed due to decay. High Rsym values are also obtained in 
cases of crystal twinning - without proper treatment- , improper space 
group assignment (although in that case, much much higher values are 
usually observed).


More information about the phenixbb mailing list