If you prefer to make your own LALSuite installation from source, make sure it is swig-enabled, contains at least the lal and lalpulsar packages, and you have run git lfs install.

Using a custom LALSuite installation requires to install PyFstat from source using the NO_LALSUITE_FROM_PYPI=1 environment variable.

First, create the conda environment provided in the lalsuite repo. Then, activate it and compile LALSuite into said environment (i.e. --prefix option in LALSuite's configure should point to the folder containing Python's bin and lib folder).

For a miniconda installation installed at /home/${USER}/opt/miniconda3, that folder would be /home/${USER}/opt/miniconda3/envs/<name-of-the-environment>. Following lalsuite instructions from their README, at the step where it tells you to run ./configure, an example minimal line would be e.g.:

./configure --prefix=`<path-to-conda-environment>`\
    --disable-all-lal\
    --enable-lal\
    --enable-lalpulsar\
    --enable-swig-python

In case of installing in different locations (i.e. running ./configure with different prefixes), make sure to run make clean once before the usual make and make install.

Second, install PyFstat from source without pulling LALSuite from a remote. To do so, clone the PyFstat repository, cd into it and, with the previously mentioned conda environment activated, run

NO_LALSUITE_FROM_PYPI=1 pip install .

Optionally, the flag -e could be added to get an editable install.