This release has been updated to handle the most recent versions of RDKit, Open Babel, and OEChem/OEGraphSim. This includes detection of the new MACCS versions they all implement (key 44 was missing), and support for RDKit’s Avalon and pattern fingerprints.
There is some general performance tuning. The FPS parser is about 40% faster and the MACCS search is about 30% faster. In addition, the NxM k-nearest search is now parallelized. Somehow I left it out when I added OpenMP support in the 1.1 release.
There are also new APIs, like the fingerprint writer, substructure fingerprint screen search and SciPy sparse matrix support, as well as improved APIs, like the context manager for the FPS reader.
Finally there’s a lot of small changes, like improved docstrings, that I hope you’ll find useful.
Chemfp has two development tracks, no-cost and commercial. The commercial version includes support for more than 4GB of fingerprint data, the FPB binary fingerprint file format for fast loading, and support for both Python 2.7 and Python 3.5 and later. Version 3.1 of the commercial track was also released today.