The LabelHash Server

Changes in LabelHash 1.0.1 (May 5, 2010):

Initial release with many improvements since the publication of our CSB 2008 paper:

Motifs can be matched against an arbitrary subset of the targets stored inside a LabelHash table. This is useful if you only want to match a motif against, e.g., a protein family.
HDF5 is the now the underlying storage format. Parallel HDF5 is used to access LabelHash files efficiently in parallel across many cores.
Boost is used heavily throughout to reduce and simplify the code base. This should make much of the code also more portable.
New, super-fast LRSMD routine, based on the Quaternion Characteristic Polynomial method.
Nearest-neighbor computation during match augmentation is sped up with pre-computed information stored in the LabelHash table.
Much of the LabelHash functionality is now accessible through a Python module.
The ViewMatch plugin for Chimera uses the new Python module, and tries to find PDB files using the Chimera preferences.
Improved software infrastructure for building, testing, and packaging.
Documentation for the web server, binary releases, and source code releases is generated from a consolidated set of files.
Code should compile out of the box on Linux (i386, x86_64, and ppc64 architectures) with the GNU, Intel, and IBM compilers. On OS X, LabelHash can easily be built as universal binaries for OS X 10.5 and above using gcc (through XCode or Makefiles, depending on how CMake was run).