Uppsala Software News @ 960806
                ******************************


* XPLO2D Version 2.0
********
The automatic X-PLOR dictionary generation facility has been
improved *substantially*, to the point where it is almost
fail-proof (note the "almost" ;-).  Among the changes:
- cleverer method to figure out which chemical element each
  atom belongs to (works even with carbon names like "2C5'")
- if your PDB file contains the formula of the compound, this
  becomes even cleverer.  For instance, an atom called "AC3*"
  would normally be considered to be actinium ... However, if
  the formula is provided, the program can figure out if there
  are any actinium atoms in the compound.  If not, it will try
  other heuristics and finally decide that this is probably an
  ordinary carbon atom
- automatic "guestimate" of the number of hydrogen atoms
  attached to each C, N and O (works very well, provided
  the geometry of the compound is reasonable, i.e. not
  if your single C-C bonds are 1.4 A)
- automatic generation of a new PDB file which contains
  the atom type as the B-factor, and the number of hydrogens
  as the occupancy.  If the generated dictionaries still
  contain errors, this file can be edited to remove the
  errors and used as input in a re-run of the program
- much cleverer algorithm for performing the atom typing,
  i.e. figuring out which atoms are equivalent
- automatic check to see if all atom type names are unique
  4-character strings; if not, the program does its best
  to fix it
- MASSes are now always correct, provided the chemical element
  and number of hydrogens were correct
- if problems or ambiguities remain, the program will print
  warning messages
- correct averaging of dihedrals and impropers (i.e., using
  RTODEG*ATAN2(AVESIN,AVECOS) in Fortran)
- carbon atoms with unusual impropers (not near zero or +/-35
  degrees) are flagged with a warning
- slightly improved X-PLOR minimisation input file
  All in all, the program now produces correct topology and
parameter files in most cases.  For example, generating the
dictionaries for a nogalamycin-derivative (57 atoms) takes
a minute or so, and the minimisation in X-PLOR another
5 or 10 CPU seconds.  If type conflicts occur, you can quickly
edit the "pseudo-PDB" file (with types as B-factors etc.),
and re-run the program, telling it to use the B-factors as
the atom type designation.
  I have also re-run the X-PLOR dictionary generation for more
than 900 hetero compounds, the result of which is available from
the X-PLOR web site at URL:
      http://xplor.csb.yale.edu/hetero/
  The manual for the program (at least, for this option) has
been completely rewritten.  It also includes a section on
"interpreting warning messages" (since I found that some
warnings are due to errors in the structure rather than to
shortcomings of the program ...).  Check it out on:
      http://onyx.bmc.uu.se/~gerard/manuals/xplo2d_man.html


* MOLEMAN2 Version 1.1
**********
The PRotein CA_analysis command now looks for sequential stretches
with "many" CA-geometry outliers.  This may be useful to detect
poor bits of structure (including out-of-register errors).
  The ONo LS_plane command has been vastly improved (this command
generates an ODL file to draw the least-squares plane through a
set of selected atoms, e.g., a ligand).  You now get nice planes
(before you could get 400 A long planes through a glucose if it
was oriented unfavourably ...).  Also, you can now give the graphics
object name and colour as parameters, as well as the size of the
border around the atoms (typically, 1 A is aesthetically pleasing).
We've used this the other day to visualise a twisting and turning
model of cellulose inside the tunnel of CBH I, which looked cute.
  The MUlti_geometry option now correctly averages dihedrals, i.e.,
using RTODEG*ATAN2(AVESIN,AVECOS).


* LSQMAN Version 4.7.2
********
The multiple-model Ramachandran plot (MRama command) now uses our
new definition of the core regions.
  All dihedrals (phi, psi, chi1, chi2) are now averaged properly,
i.e., using RTODEG*ATAN2(AVESIN,AVECOS) in Fortran.
  Bug fixed (in EXplicit) which sometimes gave a core dump on Alphas.


* OOPS Version 4.5.1
******
When checking for Ramachandran outliers it now uses our new
definition of the core regions.


* ESSENS Version 1.1.2
********
Bug fix as announced previously on O-info.  Also, the default for
"K" is now really 60% of the number of atoms, as advertised ;-)


* SEATTLE
*********
Off to Seattle tomorrow.  I'll bring some ESSENS maps to demo in
the DEC booth at some stage.
  After Seattle, I have three more trips until mid-October, so don't
expect the next Uppsala Software News any time soon.


* PLATFORMS
***********
I have had various requests for implementations of, in particular,
the X-UTIL programs ("the trivial jiffies", Alwyn would call them,
i.e. MOLEMAN2, XPLO2D etc.), for IRIX 6.2, HP, OpenVMS and Linux.
Will look into this later this Autumn after the traveling is over.
Mails with suggestion or requests can be sent to me directly,
*NOT* to the O-info list.


* SPACEGROUPS
*************
Ralf Grosse Kunstleve at Yale has collected some pages with information
about spacegroup frequencies from the PDB.  Check out:
      http://xplor.csb.yale.edu/dept-store.html
and find out that there are only 7 of the 65 chiral spacegroups left
which have not been observed for macromolecules yet (P222, P312, P23,
F23, P432, F4132 and P4332).


--cd