1. Scaling Native data to derivative data
Each of the three datasets that you have just downloaded is on a completely arbitrary scale. If these data were on an absolute scale, normalized for the number of electrons in the unit cell of the crystal, it might be possible to just throw the three of these together and compare them without scaling. However this is never safe in practice. Rather, the data first need to be placed into a single file in which all the data are forced to be in the same asymmetric unit (this is best done with the CCP4 program called cad) and then the data must be scaled (using the CCP4 program scaleit). You can download my input files that do each of these things here: cad.csh scaleit.csh
To run these programs, you must do the following:
a. Ensure the shell-script file is executable by typing (eg):
% chmod +x cad.csh
b. Run the program and capture the output into a log file using the tee command:
% cad.csh |tee cad.html
% scaleit.csh |tee scaleit.html
c. read the log file using your favorite browser:
% mozilla cad.html
or on OS X you can use
% open cad.html
to open the file in your default browser. (Omniweb handles the java scripts well.)
This allows you to browse the log files, which are written out in html code, to see what you have done.
The output of your scaleit.html file should look like this. It will include the following lines:
----------------------------------------------------------
----------------------------------------------------------
Isomorphous Differences
Derivative title: FP= Fnomnph5 FPH= F123 SIGF123 DANO123 SIGDANO123
Differences greater than 3.9538 * RMSDIF are unlikely,
ie acceptable differences are less than 42.09
Maximum difference 72.00
Difference interval for Histogram 5
----------------------------------------------------------
Anomalous Differences
Differences greater than 3.9071 * RMSDIF are unlikely,
ie acceptable differences are less than 25.90
Maximum difference 37.00
Difference interval for Histogram 3
----------------------------------------------------------
----------------------------------------------------------
Isomorphous Differences
Derivative title: FP= Fnomnph5 FPH= F124 SIGF124 DANO124 SIGDANO124
Differences greater than 3.9545 * RMSDIF are unlikely,
ie acceptable differences are less than 38.92
Maximum difference 117.00
Difference interval for Histogram 5
----------------------------------------------------------
Anomalous Differences
Differences greater than 3.8907 * RMSDIF are unlikely,
ie acceptable differences are less than 27.70
Maximum difference 153.00
Difference interval for Histogram 4
There are four entries, two for each derivative, corresponding to isomorphous and anomalous differences. I have highlighted the most important part in red (in your html log file, all of this will be in red). Those numbers represent the maximum differences that are likely to be real. The ones that are bigger are statistical outliers and in subsequent calculations we must tell the programs to ignore these or the quality of the patterson and electron density maps will really suck. So you must always look for those numbers and write them down.
The other important numbers to look for are the scale factors. Good derivatives will have isomorphous differences of around 20%. For each of the derivatives, you will find two java applet graphs in the html log file. Go to the second table and look at the scaling Rfac vs. Resolution. It goes up at highest resolution, but look at the overall statistics, printed as one line below the applet. For derivative 123 it looks like this:
THE TOTALS 6141. 2407. 0.995 0.981 0.214 0.296 0.199 8.1 72.
The two numbers in red are the scaling R-factor and weighted scaling R-factor, and they are both around 20%. For derivative 124 we have the following:
THE TOTALS 6159. 2431. 0.990 0.984 0.190 0.270 0.169 7.3 117. 4713. 0.9
Again, the scaling R-factor is close to 20%.
(A peculiarity of trigonal space groups is that a* and b* are the same length but they are not physically equivalent. For that reason there is a 50% chance that one crystal will be mis-indexed relative to the one it is being compared to. In our case, Derivative 123 was originally mis-indexed, and I had to reindex it. That is why it has the prefix R in front of it. The tip-off was that the scaling R-factor was around 60% instead of 20%. As advertised, every crystal structure solution has some unique twists.)
Back to the top.
Next: Finding Heavy Atom Derivative sites.