How to get chemfp

Chemfp has two development tracks. Nearly all development is on the 4.x track. This version has the highest performance and most features. It requires Python 3.8 or later.

The 1.x track is maintained in order to provide a good reference baseline to evaluate similarity search performance. It is available at no cost under the open source MIT license. It is only available for Python 2.7.

Version 4.1

Chemfp version 4.1, released 17 May 2023, is the most recent version of the 4.x track. See its documentation to learn more about this version of chemfp.

A pre-compiled binary package which works on most Linux-based operating systems can be installed using the following command:

python -m pip install chemfp -i https://chemfp.com/packages/

If you get the message:

ERROR: Could not find a version that satisfies the requirement chemfp (from versions: none)
ERROR: No matching distribution found for chemfp

then that's likely because you are using a Mac or other non-Linux-based operating system where I haven't built pre-compiled packages.

You may also download the files directly from the chemfp package manger download page.

Several chemfp features are either limited or disabled in the pre-compiled package. It may not be used to:

For details see the Chemfp Base License Agreement.

These features may be fully enabled using a time-based license key.

Request a license key

Short-term license keys are available for evaluation and benchmarking. Longer term license keys under licenses for academic and commercial use are available. View the licensing page to see the different licensing options and to request a license key.

Source code licensing

Chemfp is also available with source code access. This option has several advantages over the pre-compiled binary:

Visit the licensing page to see the different licensing options and to request a license key.

Version 1.6.1

Chemfp version 1.6.1 is the most recent version of the 1.x line. It is available at no cost under the MIT license.

It can be installed from PyPI using:

# NOTE: likely only useful for benchmarking and legacy purposes.
python2.7 -m pip install chemfp

or downloaded directly

The 1.x track is in maintainence mode. It does not support recent (2019 or later) versions of RDKit, OEChem, or Open Babel. It exists primarily for benchmarking purposes.

See the chemfp 1.6.1 documentation to learn more about this version of chemfp.

Some of the many improvements in the chemfp 4.x track which are not available in chemfp 1.x are: the FPB format for fast loading, new APIs to make it easier to do web development, support for more than 4GB of fingerprint data, even faster search performance, Tversky search, diversity picking, Butina clustering, and many API additions for tool development and notebook and interactive use.