Author: Luca Marconato, Ilia Kats
We will work in Python, so you should have a Python environment and ideally Jupyter Notebook or Jupyter Lab installed. On Windows and Mac, a Conda environment is probably the easiest option, while on Linux, we found plain virtualenvs (or even no virtual environment at all) to work best.
You will need the following packages from your repository of choice (PyPi/Conda/your Linux distribution's packages):
numpyscipypandasmatplotlibseabornjupyteranndatanapariPyQt5mofapy2scanpymuoncell2location
You will also need the development versions of some packages.
These need to be installed with pip/pip3, even in a Conda environment, using pip install git+repository_url
SpatialDE: https://github.com/PMBio/SpatialDEmofax: https://github.com/bioFAM/mofaxspatialdata: https://github.com/scverse/spatialdataspatialdata-io: https://github.com/scverse/spatialdata-iospatialdata-plot: https://github.com/scverse/spatialdata-plotnapari-spatialdata: https://github.com/scverse/napari-spatialdata with thespatialdatabranch. Usepip install git+https://github.com/scverse/napari-spatialdata@spatialdatanapari-matplotlib: https://github.com/matplotlib/napari-matplotlib
If you're using Python 3.11, you will also need to either install napari from git or manually apply this patch to your napari installation.
Either miniconda or anaconda is required, with Python >= 3.7.
If you didn't just install the latest version of conda you may encounter errors.
In such a case please update miniconda/anaconda (e.g. by running conda update --all) or install the latest version.
Open a terminal and create a conda environment with all the required libraries by running conda create -n day4 python==3.10 numpy scipy pandas matplotlib seaborn anndata napari notebook scanpy -c anaconda -c conda-forge -c bioconda.
If you encounter errors you can try
- Updading conda and then trying again the command above; to update conda use
- From the message error that you get, find the package that is issuing the error, remove that package from the command line shown above and run again the command. When you don't get any more errors install the left-out packages with pip (the specific pip command is package dependent, please search online)
Finally, activate the conda environment with conda activate day4 and install additional packages with pip as described above.
A trick to speed up the creating of conda environment and the installation of conda packages is to use mamba for the those tasks. Then you can continue using conda normally for activating the env, deleting it, etc..
Run this from your base conda environment for installing mamba:
conda install -c conda-forge mamba
Then run the create command by replacing conda with mamba:
mamba create -n day4 python==3.10 numpy scipy pandas matplotlib seaborn anndata napari notebook scanpy -c anaconda -c conda-forge -c bioconda
Proceed as normally.
conda activate day4
First, install the virtualenv package globally.
You may do so with pip or from your Linux distribution's repositories.
Create a virtual environment with virtualenv --system-site-packages day4.
This will create a virtual environment that will use your system's default Python interpreter and also have access to globally installed Python packages.
Activate the environment with source day4/bin/activate.
You can now install packages into the environment with pip install package_name.
If Jupyter is installed globally (e.g. from your distribution's repositories), you need to enable it to see your new virtual environment.
Run python -m ipykernel install --user --name=practical from within the virtual environment to do that.
Use pip install package_name.
This will install packages into your user directory, but without isolating different projects from each other.