Repo structure

Either use as library repo with with <package_name> and docs (and maybe notebook) dirs. Or as data analysis repo with scripts, notebooks, data, steps, results, figuresdirs.

Conda environment

Create empty environment:

conda create -n <name>

Activate environment

conda activate <name>

Add channels first to make the appear in the exported environment: Add any channels first to make the appear in the exported environment (you can always add more later):

conda config --env --append channels conda-forge
conda config --env --append channels bioconda
conda config --env --append channels etetoolkit
conda config --env --append channels gwforg
conda config --env --append channels kaspermunch

Install a subset of packages:

conda install jupyterlab pandas seaborn ...
conda install gwf jupyterlab pandas seaborn ipympl statsmodels bioconda pyfaidx tskit ete3 scikit-allel samtools bamtools vcftools

Export it to the binder folder:

conda env export --from-history > binder/environment.yml

List revisions to envirionment:

conda list --revisions

Restore a past revision:

conda install --revision 2

Once you land on stable environment this can make it more stable:

conda env export --from-history > binder/environment.yml
conda deactivate <name>
conda env remove -n <name>
conda env create -n <name> -f binder/environment.yml

Create PyPi and Conda packages

conda install conda-build twine anaconda-client

bash pypi.sh

bash conda.sh [-c channel -c channel ...]

Using submodules with Git

Read on here