CLI Reference#
This section documents the command line interface (CLI) of CHAMOIS.
Note
When installing CHAMOIS with pip, an executable named chamois will
be created in /usr/bin or $HOME/.local/bin. If the install path is
not in your $PATH, you can also invoke the command line as a
Python module with:
$ python -m chamois.cli ...
Inference#
These sub-commands allow using the inference mechanism of CHAMOIS.
chamois predict is the basic entry-point to the CHAMOIS prediction
method, computing ChemOnt class predictions from one or more BGC in
GenBank format. chamois render can be used to render class predictions
stored in HDF5 as a tree in the terminal.
Training#
These sub-commands can be used for training and evaluating CHAMOIS.
chamois annotate can be used to annotate the features of a set of
BGCs in a GenBank file, and create a features.hdf5 file suitable
for training. chamois train can to train CHAMOIS from a dataset.
chamois validate can check the performance of a trained model on
a given dataset. chamois cv can run (stratified grouped) cross-validation
to evaluate generalization.
Note
Some of these commands have additional dependencies, such as
chamois train which requires scikit-learn
to train the logistic regression classifiers. To install the
required dependencies, make sure to install the train extra
(e.g. pip install "chamois-tool[train]").
Compound Search#
These sub-commands can be used to explore CHAMOIS predictions.
chamois gompare can be used to search which BGC of a dataset is most likely
to produce a query metabolite. chamois search can be used to search which
metabolite of a compound catalog (such as NPAtlas)
is most similar to the predictions.
Model Interpretation#
These sub-commands help interpreting the CHAMOIS model.
Model Interpretation