Download

CAVER (command line version)

About CAVER 3.03 BETA

• This version uses additively weighted Voronoi diagram. It is possible to return to the previously used ordinary Voronoi diagram by the parameter “awvd no”.
• Memory requirements and computational times have significantly decreased. This can be useful for the analysis of large structures, e.g., hundred thousands of atoms.
• Utilization of additively weighted Voronoi diagram did not change significantly the properties of tunnel clusters in molecular dynamics trajectory of haloalkane dehalogenase DhaA. However, the properties and ordering of individual tunnels in static structures may change, which is due to a principal limitation of geometry based analysis of tunnels in static structures. If your conclusions are sensitive to small changes in tunnel radii or lengths, we strongly recommend to analyse a molecular dynamics trajectory rather than a single static structure.
• A new parameter was added: parallel yes|no. Voronoi diagram computation uses multiple processor cores, unless this parameter is set to no. Note that Java may still use multiple cores, e.g., for garbage collection, which has to be solved on the operating system level. In Linux, taskset can be used for this purpose. For example the command taskset -c 0,3,4 java -jar caver.jar -<other arguments> uses only CPU cores 0, 3 and 4. Scripts in the directory caver_3.0.3/caver/scripts shows automated solution for the PBS environment.

The main difference between CAVER 3.01 and CAVER 3.02 versions is the following:

• More efficient version of UPGMA clustering algorithm is available and used as default. Previous version can still be used by setting 'clustering_algorithm' to 'loewenstein' (default is the efficient algorithm - 'murtagh'). Disc space is no longer needed to store distance matrix. Clustering speed benefits from large RAM (and accordingly set java -Xmx values).

The main differences between CAVER 3.0 and CAVER 3.01 versions are the following:

• The starting point optimization procedure was corrected. Previously, the starting vertex was sometimes identified in a too narrow cavity. The change can lead to minor changes in the geometric properties of tunnels, similarly as if the parameter seed is modified.
• Three letters were originally expected for the residue name while reading the PDB files (columns 18 - 20). This complies with the PDB file Format Guide, however, some structures use four letters (columns 18 - 21). CAVER now reads four letters as default, but it is possible to return to the three letter mode by setting ‘long_residue_names’ to ‘no’.
• The atom serial number can be newly composed of the letters as well as the digits. In such cases, letters are interpreted as additional digits 10, 11, 12, etc. This is useful when loading large PDB files.

Older versions of CAVER are available upon request.

Data