xheavy



NAME

     xheavy - refine heavy atom derivatives and calculate phases.



USAGE

     xheavy [file.sol file.phs]



DESCRIPTION

     xheavy is used to calculate Multiple Isomorphous Replacement
     phases  and to refine heavy atom solutions.  The MIR phasing
     is done in an iterative manner where  SIR  phases  for  each
     derivative  are  first  calculated.  These are combined, new
     errors are estimated based on the combined SIR  phases,  and
     finally,  new  phases  are output using the new E estimates.
     The  refinement  method  used  is  a  recursive  correlation
     search.   Each  heavy  atom  is moved in turn on a finer and
     finer grid until no improvement in the  correlation  between
     the observed and calculated difference is found and then the
     relative occupancies are  refined.   The  B-values  are  not
     currently  refined. Residual maps and difference maps can be
     calculated to locate more heavy atom sites.

     There are two windows for xheavy; the main one and  an  edit
     pop-up.    On  the main window is the Crystal field which is
     the key to the file containing this  crystal  types  parame-
     ters.  Unit Cell is used as a verification that you have the
     information for the crystal loaded.   Do not edit this  here
     but  edit  the crystal file instead with xtalmgr.  Directory
     is where the program looks for files.  The  Derivative  File
     is  a  solution  file  where  the  heavy atom parameters are
     saved.   You can Load and Save  the  solutions  files.   The
     save  button  is  a menu button that gives you the option of
     saving a single derivative (the one that is  highlighted  in
     list), all of them, or outputting all of them in a PHASIT (a
     Bill Furey program from his  PHASES  package)  format  file.
     Several  derivatives  can  be  put into a single solution by
     loading their individual solution files and then saving  all
     of  them in a new file.  Output Phases is the name of a file
     that will contain the output phases.   Next is the  list  of
     Derivatives  where  a  single  derivative can be selected by
     clicking on and highlighting it.   Below are  three  buttons
     for  manipulating  the  list.   Use New to start a list from
     scratch.  Edit is used to enter/replace parameters and sites
     for  the selected derivative.  Selecting it pops up the edit
     window (see below) and saves the current data in  a  buffer.
     Delete deletes the currently selected derivative.

     With Method you can choose, in the order they are  meant  to
     be  used,  either  Correlation Refinement, Calculate Protein
     Phases,  Map  coefficients  or  Clear  Phases.   Correlation
     Refinement  is  unique  to  this program but has proved very
     robust and successful.  It is independent of scale factor so
     that it is not necessary to start with a close approximation
     of the occupancy as in least-squares methods.  Since it is a
     search  method,  it  finds  the  global  minimum (within the
     search radius) and does not get stuck in  a  local  minimum.
     (For  the  true  global minimum use xhercules which searches
     the entire unit cell using the same  correlation  function).
     Protein  phases  are  calulated in a several step procedure.
     First, the single isomorphous replacement  phases  are  com-
     puted for each derivative.  These are then gathered together
     to get a rough estimate of the protein phase.   Fc  is  then
     rescaled  using  the  protein phase to account for the extra
     scattering power of the heavy atom (i.e. the scale should be
     greater than 1.0 - for a well-occupied derivative it is from
     1.05 to 1.10).  Better error coefficients  are  then  calcu-
     lated  and  these  are used to rephase each derivative; then
     the final protein  phases  are  gathered  from  these.   The
     method  is  similiar  to  lack-of-closure  error  refinement
     except that the only parameters changed are the  errors  and
     the  scales. The map coefficients should be calculated for a
     selected derivative after the phases are  calculated.    The
     map  coefficients  option  creates  a  phase  file  for  the
     selected derivative with the coefficients:  h,  k,  l,  |FP-
     FPH|,   (FPH-FP)  -  FHcalc,  aprotein.   Use  xfft with the
     Fourier type Fo to make a heavy atom difference map or Fo-Fc
     to  make  a  residual  map.   These are used to look for new
     sites.  Finally we come to the Apply button which is used to
     start one of the methods and Abort is used to stop it in the
     middle if something goes wrong.


  SOLUTION EDITOR
     Solutions can be read from xpatpred or entered with the edi-
     tor.   The  derivative  picked on the list of derivatives is
     edited when the Edit button is pushed.   First  the  deriva-
     tive  title  is  entered.   A  fin or df file containing the
     difference information must be specified  in  the  DataFile.
     Tell  the program which one it is with File Type.  Addition-
     ally, you can specify which of the columns in  the  df  file
     contains  the  phasing  pair.   One  of three phase types is
     specified next: isomorphous, isomorphous anamolous (must use
     a  df  file), and native anamolous.  You can downweight this
     derivative relative to other  derivatives  with  the  Weight
     slider.   There are two resolution limits between which data
     is used.  Sigma Cut is used to remove small weakly  observed
     reflections.   If  either of a pair of measurements is below
     this value, the pair is not considered.  Then comes a filter
     for  removing  ridiculous  differences  by comparing the the
     value EMBED Equation.  If this is greater than the value  in
     filter,  then the reflection is rejected.  A value of 100 is
     about right for most derivatives.   Derivatives  with  large
     differences  can  use  a  larger value of about 130.   Below
     this is the list of heavy atom Sites.  They are edited  with
     the  controls  below.  Start  the list by first entering the
     values you want and then select Insert.   Insert will insert
     at  the end of the list if nothing is selected, otherwise it
     inserts after the selection (however, never select a line in
     the  middle of entering values as it will overwrite what you
     have just done!).  To edit, first highlight the atom in  the
     list  you  want  to  change, edit the values and then select
     Replace.   To delete an atom, select it and then use Delete.
     Next  to  the  Atom  type  is a little pull-down menu button
     (with the triangle) that lists all  of  the  available  atom
     types.   It is best to use the pull-down menu and select one
     than type it in to avoid typing errors.   The choice of atom
     type is non-critical in the isomorphous and native anomalous
     case as it mainly just changes the scale factor.    It  does
     matter when using the isomorphous anomalous phasing calcula-
     tions because it sets the ratio of f'' to f'.   If your atom
     does  not exist, choose the closest one.  See below for more
     information about adding new scattering factors.   When  you
     are done editing you can either push Apply to save the edits
     or close the window.  (Although this may be  window  manager
     dependen  apply  is  safer.)   If you mess things up you can
     recover to the last saved position with Reset.

     Verbose output is given in the message window.  Don't forget
     that  this can be saved using the MENU button in this window
     to pull up the FILE submenu.


  NOTES
     This program is under active development.  As mentioned, the
     native  anomalous  phasing  has  yet  to be implemented. The
     function is there to remind the developer of  his  shortcom-
     ings and to tease users.  The program has been used to solve
     several structures.  In particular, the correlation  refine-
     ment works very well.  To date, no one has trusted it as the
     only source of phasing, however.  So far everyone  has  used
     the  phasit  output  option  to confirm the phases with that
     program.  Comparison of maps shows  very  little  difference
     between  the  phases  calculated,  but the calculation is so
     important that no one was willing to rely on the newer  pro-
     gram.    Also,  PHASIT  prints  more  statistics  which  the
     developer promises to add in  some  future  version.   Right
     now,  one  has  to  rely  on Rcentric and figure of merit as
     guide posts.  Adding new scatterers  is  cumbersome  because
     the  program  must  be  recompiled.   See the README file in
     src/xheavy for the location of added elements.



VERSION

     Release 3.2 of XtalView


SEE ALSO

     xtalview(1)