Knitwork

Refactor of FragmentKnitwork

Installation

Knitwork is published to PyPI as xchem-knitwork (the Python package is still imported as knitwork):

pip install xchem-knitwork

For development, install from a clone of the repository:

git clone https://github.com/xchem/Knitwork
cd Knitwork
pip install --user -e .

Testing

Tests use uv and the dev dependency group:

uv run --only-group dev pytest

Releasing

Releases are made by pushing a semantic version tag (with no v prefix), e.g. 1.2.0 or 1.2.0-rc.1. The package version is taken from the tag. The release workflow then publishes: -

• The Python package to PyPI (using PyPI Trusted Publishing)
• A Docker image to Docker Hub (as xchem/knitwork:<tag>)

Configuration

The configure command can be used to set variables that the package will use when running commands:

e.g.

python -m knitwork configure GRAPH_LOCATION XXXXX
python -m knitwork configure GRAPH_USERNAME XXXXX
python -m knitwork configure GRAPH_PASSWORD XXXXX

Running Fragment Knitwork

Run the following steps to generate merges from ligands in an SDF.

Fragmentation

To run the Fragment process which looks for subnodes and synthons for a given set of fragments/molecules in an SDF and groups them into pairs:

python -m knitwork fragment INPUT_SDF

This will generate pickled pandas dataframes, along with caches and other outputs in fragment_output by default:

• molecules.pkl.gz: pickled dataframe of input molecules
• molecules.sdf: SDF of input molecules
• pairs.pkl.gz: pickled dataframe of output pairs

For more options see:

python -m knitwork fragment --help

Pure Knitting

To query the graph database for "pure" merges matching fragment pairs in the fragment_output folder by default:

python -m knitwork pure-merge

This will generate pickled pandas dataframes, along with caches and other outputs in knitwork_output by default:

• pure_merges.pkl.gz: pickled dataframe of merges
• pure_merges.sdf: SDF of merges

Impure Knitting

To query the graph database for "impure" merges matching fragment pairs in the fragment_output folder by default:

python -m knitwork impure-merge

This will generate pickled pandas dataframes, along with caches and other outputs in knitwork_output by default:

• impure_merges.pkl.gz: pickled dataframe of merges
• impure_merges.sdf: SDF of merges