Introduction to deep learning for scRNA-seq

Authors: Sandeep Santhosh Kumar, Jiadong Mao, Xiaochen Zhang

Audience	Prerequisites	Duration
Anyone	Install Anaconda (including Python and Jupyter Notebook)	~ 20 min
Anyone	Install Python Packages	~ 20 min

Description

This repository includes material for a hands-on workshop 'Introduction to deep learning for scRNA-seq'. We show you how to use scvi-tools to analysis your own scRNA-seq data.

Installing Anaconda navigator to set up jupyter notebook environment

(you can skip this if you already can run jupyter notebooks in an environment)

• A recommended short 3 min youtube video on setting up anaconda environment and jupyter notebook can be found here (https://www.youtube.com/watch?v=TzUfBhpr5Kw)(Ignore the opening the terminal step and install the jupyter notebook or jupyter labs from Home as mentioned at the end of the video.
  The main steps:
  • Install Anaconda navigator, select a graphical installer
  • Create an environment. Environments on left menu-> create option at the bottom -> name the new environment and create. Python version should be 3.9 or more recent (recommended is 3.9). You can find the documentation here.
  • Install jupyter notebooks in the environment. Home on the left menu -> check if you are on the new environment, the name should be selected on the dropdown menu after on on the top of the page -> Click on install under the jupyter notebooks icon.
  • Click on launch under the jupyter notebooks icon and navigate to this workshop notebook and proceed to install the packages.

Installing scvi-tools and some other packages

Open your jupyer notebook and run these code: (Only need to run this once, installs the necessary packages, this will take afew minutes)

import sys
!{sys.executable} -m pip install scanpy scvi-tools seaborn scikit-misc igraph leidenalg

Typically, it's better practice to perform conda installations when using conda environments as we are, but if we don't have too many dependencies we get away with pip installations.

More details on the packages:

• scvi-tools
• scanpy

Material

Click here to access or download the jupyter notebook for this workshop.

Data

Datasets are from the scvi-tools package.

Solution

Click here to access the solution.