FastROCS TK 0.3.0

• Added ShapeDatabaseAddMols.py script to add molecules to a running ShapeDatabaseServer.

• Added ShapeDatabaseIsLoaded.py script to report whether the initial database load has completed.

• Added argument to ShapeDatabaseClient.py for the number of hits to return, defaults to 100.

• Redesigned how the BestShapeOverlay.py script runs to be more useful for doing evaluations.

• Restricting OEShapeDatabase to only run on one type of device and prefer to run on a GPU, and fall back to a CPU device only if there is no GPU present. Used to be that on OS X the search would run on the GPU and CPU simultaneously, however, the results of the overlay could differ depending on whether they were sent to the GPU or the CPU. Due to dynamic load balancing between the CPU and GPU this could change the rank of the results returned in an apparently non-deterministic manner. It is not pleasant for an end-user to run the same query multiple times and get different results so we restrict searches to either CPU or GPU, not both now. Note, this only would occur on OS X. The NVidia implementation doesn’t support a CPU device, so this never occurred there.