Biskit

-----------

This Biskit release was long overdue. Even though I didn't have much
time to work on Biskit for most of the last two years, a lot of
changes and improvements have accumulated, especially over the last
months. The SVN trunk is now quite different from the last
release. This release is (or should be) still backward-compatible to
binary pickled data from all previous releases. I haven't insisted any
on compatibility with the old Python 2.4 any longer. If anyone still needs
support for that, patches can be generated quite easily -- please send
me (Raik) a mail. The "classic" Python versions 2.5 upwards should
work. A conversion to Python 3.0 is planned but will require some more
effort.

Warning Windows users: The Windows installation package seems to work
but we have not developed for this platform and there are some bugs,
especially, related to the expansion of path and file
names. Surprisingly enough, many things seem to work. I even received
a report of the homology modeling pipeline running on Windows.
See: http://biskit.pasteur.fr/install/windows_install

The most important new features:

* TM-align: structure alignment wrapper and integration into PDBModel
- new class: TMAlign
- new method in PDBModel: structureFit()

* Reduce: a wrapper for the reduce program to optimize hydrogen
bonding networks, assign protonation states to His and add hydrogens
to a structure.

* AtomCharger: maps partial atomic charges from Amber prep/topology
files into the residues of a given structure.

* Delphi: wrapper (and more) for automatted calculations of
electrostatic potentials with the DelPhi Poisson-Boltzmann solver

* DelphiBindingEnergy: workflow for the calculation of electrostatic
free energies of binding (with Delphi) for a given complex

* Support for DNA and RNA residues in entropy and other calculations

* BioModel: extract and apply the biological relevant unit from PDB
file (by Alexander Gryzlov)


Smaller improvements:

* new plotting methods for ProfileCollection (e.g. histograms)

* DNA/RNA support in PDBModel.maskBB() (backbone)

* also report PDB chain IDs in PDBModel.report()

* new PDBModel.reportAtoms() to print PDB-style info for selected
atoms

* new PDBModel.unequalAtoms() tells which atoms are *not* matching
between two structures (opposite of compareAtoms() )

* improved support for empty PDBModels, e.g. concatenation to and from

* re-written and simplified the recognition of sequence repeats
that is used by PDBModel.compareAtoms() (attention, this may cause minor
differences in results from this method)

* Executor: allow bundling of temporary files into temporary directory

* changed identification of chain breaks

* additional test cases

* updated Amber related tools (AmberParmBuilder, entropy-related classes)
to Amber version 11

* Biskit.Mod blast searches adapted to recent BioPython / NCBItools versions

* Biskit.Mod.SequenceSearcher can now fetch fasta records from NCBI
website This allows to run homology modeling without a local
database installation but isn't yet passed through as an option to
the end-user scripts.

* Removed the warning message for when biggles or PVM are not
installed; Errors will instead be raised when these modules are needed.


... and many fixes for old bugs without, hopefully, introducing too
many new ones.


Unfortunately, there are also some things that will NOT work right now:

* the new version of HMMer and mofifications to Pfam have killed the
Biskit wrapper which probably would need to be re-written almost
from scratch -- the PDBDope.addConservation doesn't currently work

* the Fold-X wrapper is very outdated and has not been updated to the
latest versions

* PSI-blast searching may not be working with the latest NCBITools/BioPython
version. The Mod module needs to be updated to the new blast+ tool
set.

See also the list of current bugs (and report new ones) at:
http://sourceforge.net/tracker/?group_id=150896&atid=779191

As always, feel free to send your comments or questions!
Good luck!
Raik

-------------

Reverts setup scripts back to using distutils. The setuptools /
easyinstall version turned out to be too error prone. This means that
setup.py will not any longer attempt to install missing python
dependencies. You have to install numpy, scientific python, and
[optional but useful] biggles, BioPython, and scipy yourself.

The setuptools version setup_ez.py is still used to build Debian
packages.

be installed system-wide with the standard 'python setup.py install'
command. This required a change to Biskit's directory layout: The
'test' and the 'external' folder were moved from the project root into
the Biskit python package. That means:

old project new project
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
* biskit/external -> biskit/Biskit/data
* biskit/test -> biskit/Biskit/testdata
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

The idea here was to have all extra folders nicely bundled with the
python package rather than spreading them accross your system. This
also applies to the scripts and docs folder. These two remain at the
project root, but are copied into the package folder during
installation:

svn project after installation in site-packages
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
* biskit/Biskit -> site-packages/Biskit
* biskit/scripts -> site-pacakges/Biskit/scripts
* biskit/docs -> site-packages/Biskit/docs
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

The new setup.py will attempt to automatically fetch, compile and
install missing Python modules (numpy, ScientificPython, scipy,
biggles, BioPython). That means, in the ideal case, you can install
Biskit and all dependencies with a single command.

However, the compilation of these modules from source depends on
non-Python C libraries like netcdf (for ScientificPython) and
plotutils (for biggles) and it requires that your system has the
standard make and compiler tools as well as Python development files
(headers) installed. You may thus be better off to install these
dependencies beforehand from pre-compiled packages.

Another advantage of the new setup is that we can now start creating
Debian, RPM and even Windows installation packages -- these should be
considered experimental. Please report success or failures!

Other changes:

* fixed Intervor wrapper

* Python 2.6 compatibility

* no need of installing the old Numeric next to the new numpy

* remove last depencies on old Numeric package from test data

* some minor bug fixes

* some work-in-progress packages

previous beta release:

* switch from CVS to SVN version control

* PDBModel: new report() and plot() function to get a fast overview
over the content of a PDB

* PDBModel: PDBModel( '3TGI' ) will fetch the PDB entry 3TGI directly
from the ncbi database or a local mirror

* PDBModel: new maskDNA() function

* new CommandLine.py is a candidate to replace the old tools.cmdDict command
line option parsing for scripts; with type-casting and test fixture
built in

* Statistics/ROCalyzer module calculates ROC areas and their statistic
significance

* new Intervor module wraps the Voronoi interface calculator of Cazal et
al (but Frederic hasn't got around to offer a intervor version for
download yet... come on Frederic ;o)! )

* PDBModel overhaul -- the atom dictionaries are gone and all the
information is now unified into `atoms` and `residues`
ProfileCollections, take, compress etc. are **much** more efficient,
profiles can be accessed in a more intuitive way, chain borders are
tracked more consistently, ... The new PDBModel is, as usual,
backward-compatible to pickles of previous versions.

* Migration from Numeric to Numpy -- plenty of bugs in the old Numeric
have forced us to migrate earlier than anticipated. Unfortunately
some third-party libraries like Scientific and Biopython still
depend on Numeric -- so both libraries need to be installed. Also we
needed to convert the biggles module to the new numpy library.

All tests I can currently perform (this excludes PVM, homology
modeling and some interfaces to external programs) are running through
without problems. But since the changes are very widespread and at the
center of the package, I expect that there are some bugs still
awaiting discovery. Furthermore, I haven't yet converted the scripts
folder to the new numpy and, since we are right now lacking a good
test suite for this part of Biskit, many of the biskit/scripts will
probably be broken (albeit not difficult to fix).

For this reason, we label this one as a "beta" release. Users who are
mostly interested in reproducing one of the existing workflows
(docking or homology modeling scripts) should for the moment stick to
release 2.0.1. However, for everyone else, the improvements should
outweight potential bugs and you spare yourself a later
migration. Please don't hesitate to report any problems!

------------------