EMAN

Substantial improvements were made in refine2d.py. Using refine2d.py on all new data sets is strongly suggested. Some programs use the EMAN2 style of arguments, i.e. "program <file> --option=value", rather than the old style, "program <file> option=value". There is a new HDF5 format compatible with EMAN2. New AIRS programs were added, such as "skeleton". New options were added to make refinement work better on large icosahedral objects. The parallelism infrastructure was improved, though network-related problems may still exist. Random model generation in makeinitialmodel.py was fixed.

A major overhaul of the parallelism infrastructure
(runpar) was done. It now uses fileserver for both
reads and writes in the cluster versions. A binary
release was made for AMD64, and support for OSX
was improved. A new program, refine2d.py, was
added for generating reference-free class-averages
from a set of particles. A new program,
makeinitialmodel.py, was added for constructing 3D
models from blobs. The AIRS software was greatly
expanded and improved with Chimera bindings. Major
improvements were done to the experimental 2D
crystallography preprocessing program (qindex).

The dfilt option now works properly. Support for
icosahedral particles was improved. MRC/CCP4
compliance was improved. The new "qplot" program
was added for quick plots of 2D data. support for
files larger than 2G was fixed. Improvements were
made to the EMANimator animation module for
Chimera. Support for Opterons was improved.

The "dfilt" option was added in refine. A
prototype of an animation module and other tools
for Chimera were created. Chimera output was added
in segment3d. Autosegmentation of symmetric
structures is now done. A bug in PNG image writing
was fixed. PGM image writing was improved.
Chimera-compatible MRC header information is put
on all MRC file writes. Improvements were made to
procpdb.py. Bugs with alignments in images not
divisible by 4 were fixed. Binaries for more Linux
platforms are now available.

This version features better python bindings, major
improvements to parallelism, persistent sockets, new
image similarity criteria, more accurate alignments, soft
edges on automasks, and an option for applying known
structure factor for amplitude correction during
refinement