Last week you asked Arp/wARP to automatically trace the
electron density map you calculated for proteinase K.
The Rwork=15.1%, Rfree=19.2%.
This model is fairly complete except for some odds and ends.

Today you will be using this arp model to begin refinement
of the prok-pmsf complex.  PMSF is the inhibitor which 
covalently bonds to the active site serine of PROK.

We begin by calculating a difference Fourier map
in order to see what differences exist between the
two structures.  Hopefully there will only be a few
changes.  The primary change should be the addition
of PMSF.
Fobs(PMSF)-Fcalc(native) phi(native)

This only works if the structures are isomorphous.
If the two data sets had very different unit cells
then this method would not work so well.

We will get positive and negative peaks in this map.
Positive peaks  in this map should correspond to atoms in 
the PMSF complex that were absent in the native protein.

Negative peaks in this map should correspond to atoms in 
the native complex that are absent in the PMSF complex

You will alter the arp/warp model to make it fit the
difference maps.  Start with the highest features first.

For example:
1) flip a tryptophan ring,

2) build a cis peptide that was absent

3) build in a PMSF inhibitor.

4) there will be lots of ordered water molecules
that you will not be modeling on this round.

When you are done modifying the coordinate set, you will
write out coordinates.  This will be your first model
representing the ProK-PMSF complex.

You submit
the coordinates to refinement against the PROK-PMSF structure
factors.  The refinement program, refmac, will move the xyz
coordinates and adjust temperature factors (4 parameters
for each atom) to minimize the
discrepancy between Fobs(Prok-PMSF) and Fcalc(Prok-PMSF).
Hence, the working Rfactor will decrease.  Hopefully the
Free Rfactor will also decrease.

A new difference Fourier map is calculated Fobs-Fcalc,
that will hopefully have less features than the previous
difference map.  If there is time, we can check this
map.


We will be reporting the statistics of the refinement
in a table, which you will hand in at the end of class.

In addition to statistics regarding structure factors 
(# of reflections used, Rwork, Rfree, etc).
You will be reporting statistics regarding the quality
of the geometry of the model. RMS deviations from ideal bond lengths 
bond angles and # of outliers from the Ramachandran plot.
The statistics you get will be rather good because of
the quality of the data.  You can use these a standard
when judging other papers.

In addition, there are two other structure validation
programs that you can use to judge the quality of the
model.  They analyze parameters that are not normally
restrained in the refinement procedure.  They look
at atom environements and how representative these
environments correlate from a library of atom environements
derived from a collection of high qualtity structures.

Mike