2.  Finding Heavy Atom Sites

This uses three input files for cns.  The first of these reads in the (in this case) two hkl data files and automatically scales the data and calculates a patterson map. The second file reads in the data and tries to solve for the heavy atom sites.  The third takes a trial site and calculates the locations of peaks in the patterson map so that you can check to see if the solution matches what you observe.  In addition, we will use a cns utility called plot_patterson.

All of these examples are based on the CNS turorial, which you should have a close look at.

a.  Calculate a patterson map.

Download the file patterson_map.inp


Edit the file using the cns_edit utility.  You have to be connected to the internet to use this.  I've already included the correct enteries, but you could have also started using the generic input files from the web site.  The following syntax invokes the editor:

% cns_edit patterson_map.inp

Run it:

% cns < patterson_map.inp |tee patterson.log

When you are done you will have a patterson map called

patterson_map_z.map

You can plot a harker section by doing this:

% plot_patterson
patterson map file -> patterson_map_z.map
patterson plane (x/y/z) -> z
postscript plot file [patterson.ps]->
minimum contour level [2.0]->
contour increment [0.5]->
plot scale factor [0.06]->
plot title -> isomorphous patterson map

If you are using my cns OS X installation, the postscript file is automatically converted to a pdf and is displayed with preview.  The peaks are pretty obvious.

b.  Solve for the heavy atom site.

There is only one heavy atom, a Br we stuck in on Stem I when we synthesized the RNA.  This is an ideal case, because we know how many heavy atoms are bound (one per molecule) and we know the occupancy (pretty close to 1.0 on an absolute scale).  You can solve the patterson map manually by printing out the other harker sections as an option in patterson_map.inp.  However, CNS also provides an automated version in heavy_search.inp.  Again, you can edit my input file, and run it:

% cns < heavy_search.inp |tee heavy.log

The top solution is the correct one:  heavy_search.sdb   You can verify this using  predict_patterson.inp as I have done here.

% cns < predict_patterson.inp

The map returned is predict_124_iso_z.map and you can plot this using plot_patterson as was done for the original map.  Here is how the pdf should look. You can now use this solution to solve the structure.  It turns out there is another solution that is equivalent (you can verify this by predicting the patteson map with it -- you get the same peaks) but it has the merit of producing the map that has the same origin as that used originally to solve the structure that resides in the pdb.  You can download it here and call it ccp4_generated.sdb You can solve the structure using either of these, but I am going to use the one that lines up with the deposited coordinates because it will make life simpler.  Please note that this is simply for my convenience and the choice of origins is arbitrary.  If you don't believe me, it will only take you a few more minutes to solve it both ways to prove to yourself that you get the same map either way.