Preparations#

Setting up Python#

The workshop will consist of a mix of introductory talks for the individual Python packages and a number of interactive coding sessions - see the Schedule for details.

You are welcome to just cruise along but to make the most of it you will probably want to follow the coding and participate in the exercises on your own machine.

In general, we expect you to have at least basic familiarty with Python including the standard data science libraries such as numpy and pandas.

Most of the workshop will be using Jupyter for interactive data analysis. Please see their website for install instructions.

Prerequisites#

Windows, MacOS, and Linux should all work. You can also run Python in a Jupyter notebook in the cloud, using Google Colab.

The tutorials are compatible with Python version >= 3.9. You should have a current version of pip (>= 23) and setuptools (>= 68).

We will use the command python3 to invoke Python on Linux, but on Windows the appropriate command may be python. You should find out what the appropriate command is for Python on your system.

Both pip and setuptools can be updated using pip itself with

python3 -m pip install --upgrade pip setuptools

In the tutorials, we use some software which is available from PyPI but not as a conda package. If you prefer to use conda, you will have to install some things in a conda environment using a conda-managed version of pip.

In order to avoid dependency conflicts with other Python packages on your device, including packages managed by the system package manager rather than pip or conda, you should create a virtual environment to install all Python packages that will be used in the workshop. Start by navigating to the directory in which you plan to work, and follow the instructions appropriate for your package manager.

Alternatively, if you plan to use Google Colab for everything, you can skip to that section.

Dependencies#

The workshop will focus on four main packages.

In addition to these core tools, we will rely on some other software:

• scanpy, a Python package for the analysis of single-cell gene expression data
• Jupyter, an interactive environment for evaluating Python code and plotting graphs
• ipywidgets, which extends the functionality of Jupyter (e.g., progress bars)
• leidenalg, an implementation of the Leiden clustering algorithm
• umap-learn, which is useful for visualizing the clustering structure of high-dimensional data
• pandas, a dataframe library for manipulating and filtering tabular data
• plotly, a rich visualization library

Installing dependencies with Pip#

Pip is the standard package manager for Python. We recommend you use a virtual environment which you can create like so:

python3 -m venv ./neuro_workshop

After creating the virtual environment, you should activate it, which sets certain environment variables and path variables to ensure that Python commands and modules refer to code in the ./neuro_workshop folder. The way this is done depends on your operating system:

.\neuro_workshop\Scripts\activate.ps1

source ./neuro_workshop/bin/activate

You can tap out of the virtual environment with this command:

deactivate

Before continuing, check the left-hand side of your terminal prompt to ensure that the virtual environment is activated! Then run this to install all required dependencies in one go:

python3 -m pip install jupyterlab ipywidgets cajal scanpy leidenalg navis umap-learn pandas plotly pynapple nemos matplotlib requests

You should be good to go! Try firing up Jupyter lab and import e.g. navis!

Installing dependencies with Conda#

Conda is a package manager commonly used in data science and bioinformatics. We do not provide a conda package for our tools; these instructions discuss how to use conda to create a sandbox for pip. Instructions for working with virtual environments in conda can be found here. You can open a conda terminal from the Windows start menu by launching the program "Anaconda Powershell Prompt."

Create and activate a virtual environment with:

conda create --prefix ./neuro_workshop python=3.11
conda activate ./neuro_workshop

Install dependencies as follows:

conda config --append channels conda-forge
conda install numpy scipy jupyterlab ipywidgets scanpy leidenalg umap-learn pandas plotly

Now run this:

pip install cajal navis[all] pynapple nemos

and pip should install the relevant dependencies into the ./neuro_workshop directory you have just created. Try firing up Jupyter lab and import e.g. cajal!

Google Colab#

Google Colab is a cloud computing service which can execute Jupyter notebooks stored in Google Drive. To use Google Colab for this tutorial, go to https://colab.research.google.com/ and click File > New Notebook in Drive

Install the Python dependencies using ! to run a shell command from a Jupyter cell:

!pip install cajal scanpy leidenalg navis umap-learn pandas plotly pynapple nemos matplotlib requests

Data#

During the workshop, we will be analyzing data from the 2020 paper "Integrated Morphoelectric and Transcriptomic Classification of Cortical GABAergic Cells" by Gouwens, Sorensen, Baftizadeh et. al.

All artifacts for the NAVis (Monday) and CAJAL (Tuesday) presentations are available in this Dropbox folder. Just download the whole thing and put it somewhere you will be able to find it during the workshop.

For Pynapple and Nemos (Monday), the workshop tutorial discusses how to download the necessary data - see Part 1 and Part 3, in the "Data Streaming" section.

Getting Help#

The workshop organizers can be contacted for installation help in the Slack group for the workshop or using the Discussions in this website's Github repository.