🚀 go-pugleaf

				                     Version 4 of HEAVY

I am pleased to announce that version 4 of HEAVY is now available.  The 
HEAVY program contains the old "HEAVY" and "HASSP" programs in an 
integrated package that is easy to use and that runs on a variety of 
unix and VMS machines.  Some of the major new features of HEAVY are 
CORRELATED PHASING, BAYESIAN WEIGHTING, and BAYESIAN DIFFERENCE 
REFINEMENT (see below for brief descriptions of these features). 
________________________________________________________________________

HOW TO GET HEAVY version 4:  HEAVY is distributed free of charge for 
non-commercial users.  All users need to get an official of their 
institution to sign a license agreement.  To obtain a license, please 
contact Nancy Cameron at the Los Alamos Industrial Partnership Office in 
the following way:

		Email: Ncameron@lanl.gov

		1.  Specify NON-COMMERCIAL or COMMERCIAL license
		2.  If you are a University of California institution, say so
		3. Your name and email address.  If you have an anonymous ftp site, 
please list it.
		4.  Name and address of person who will sign for your institution 
(i.e., person in your business office. This person will know if they can 
sign for the institution.  Faculty  at universites cannot.)
		5.  Node name(s) of single site or cluster of computers where HEAVY 
will be used.
________________________________________________________________________

FOR FURTHER INFORMATION ON HEAVY:  contact me at "terwilliger@LANL.gov".

--Tom Terwilliger	
________________________________________________________________________

NEW FEATURES OF HEAVY:

CORRELATED PHASING allows you to make use of data from a set of 
non-isomorphous derivatives that share the same non-isomorphism.  In 
effect, correlated phasing uses the information in the differences among 
derivatives that is ignored by Blow-Crick phasing.  If there is 
correlated non-isomorphism, these differences actually contain much of 
the phasing information and using them can make a dramatic improvement 
in phasing power. 

BAYESIAN WEIGHTING macromolecular atomic refinement consists of WEIGHTS 
of the form WGT=1/(sigma**2+E**2) where the E**2 values are estimated 
from rms differences between observed and calculated amplitudes of 
structure factors. A structure factor file that can be read directly by 
X-PLOR is generated. This weighting scheme leads to lower errors in 
coordinates than unit weighting.
 
BAYESIAN DIFFERENCE REFINEMENT is a method for refining a macromolecular 
structure  (a "variant") that is very similar to a structure that has 
already been solved (the "native").  By refining differences between 
native and variant observed  structure factor amplitudes against 
differences between native and variant calculated structure factor 
amplitudes, very accurate differences between the two structures can be 
obtained. Difference refinement requires that the native and variant 
structures be similar. In Bayesian difference refinement, an optimal 
weighting scheme for combining difference refinement and conventional 
refinement is used. Bayesian difference refinement allows all the data 
for the variant structure to be used and represents a substantial 
improvement over both conventional and difference refinement. 
________________________________________________________________________