I've just released a preview release of chemfp, version 4.0b2. It includes a new "chemfp" command, diversity selection, a "high-level" API, pandas integration, progress bars, new output formats, and more.
It is ready for production use. What remains is to improve test coverage and update the documentation.
To install it on Linux-based OSes do the following:
python -m pip install chemfp==4.0b2 -i https://chemfp.com/packages/
This version is available at no cost under the Chemfp Base License Agreement. This license lets you use most chemfp features in-house, and generate FPS files for any purpose.
You will need a license key to search a collection of over 50,000 fingerprints, to do Tversky search, to generate FPB files, and to use the new diversity selection methods.
Compatibility
This release should not break any existing code that uses chemfp. The command-line tools are still there, with the same options, and the same for the API.
There is one deprecation warning. The bitops.byte_difference
function has been renamed to byte_xor.
chemfp command
What's new is the chemfp command-line tool, which accepts
subcommands like chemfp simsearch and chemfp maxmin. This gives
chemfp a way to add new tools without adding new commands to the
shell. Use chemfp --help to see what's available.
Diversity selection
Chemfp 4.0 add support for diversity selection. The three new methods are MaxMin, heapsweep, and sphere exclusion (both random and directed).
The command-line tools (simsearch, fpcat, etc.) are still there