For users

How to launch course tutorials

https://github.com/statgenetics/statgen-courses/wiki

Available tutorials

Caution: docker images listed below are obsolete as of April 2024, but kept here for backward compatiblity suport. The updated setup is a single image found at https://github.com/cumc/handson-tutorials/tree/main/setup/docker

Available tutorials (via --tutorial option of statgen-setup script) are those with docker images available as listed here.

For developers

Caution: Instructions for developers is obsolete as of April 2024, but kept here for backward compatiblity suport. The updated instructions can be found at https://github.com/cumc/handson-tutorials/blob/main/setup/docker/README.md

• docker folder contains files for docker images to run the statgen course tutorials.
• handout folder contains some handouts.
• src folder contains utility scripts eg, tools to setup the Jupyter server online.

Prepare your computer to manage the tutorials

Software you need to install on your computer are SoS (simply type pip install sos to install, or, check out here for alternative installation methods if you have troubles with that command) and docker. Additionally to run the course material on your computer (not on cloud VM) you have to put src/statgen-setup script to your PATH and change it to executable, eg, chmod +x ~/bin/statgen-setup if you put it under ~/bin which is part of your PATH. To verify your setup, type:

statgen-setup -h

you should see some meaningful output.

Build tutorial specific images

gaow/base-notebook, a minimal JupyterHub / SoS Notebook environment for scientific computing, is used to derive tutorial specific images in this folder.

To build tutorial images and push to dockerhub, eg for tutorial igv found under docker folder, please execute command below under the root of this repo (same folder as this README.md file)

statgen-setup build --tutorial igv

Or, multiple tutorials,

statgen-setup build --tutorial igv vat pseq regression annovar

You can use option --tag to add version tag to a build, eg, --tag 1.0.0.

If you run into this error denied: requested access to the resource is denied please make sure you have push access to dockerhub account statisticalgenetics. Please contact Gao Wang for the password to that account and use docker login command to login from your terminal. Then try build again.

Incorporate JupyterLab setup script

It is possible to additionally customize the docker image when started from JuptyerLab environment, to download the latest version of tutorial notes and deploy small data in the JupyterLab server launched from the docker image. To configure please study this example (which is self-explanary and I'll not elaborate it here).

If your tutorial comes with a large data-set it is not suggested that a setup script is used. Instead, you can still install the pull-tutorial.sh script and instruct users to type a line of command get-data from JupyterLab terminal when they first logged in to the server. See this example for details.

Setup course JupyterHub server on your computer

To set it up for selected tutorial(s) on your local computer, for example for vat and pseq tutorials,

statgen-setup launch --tutorial vat pseq

After all steps are complete, you check the Jupyter Hub server on your machine:

$ docker ps

CONTAINER ID        IMAGE               COMMAND                  CREATED             STATUS              PORTS                              NAMES
4888c67e9774        vat_hub_user        "tini -g -- jupyterh…"   8 seconds ago       Up 7 seconds        8888/tcp, 0.0.0.0:8847->8000/tcp   vat_hub_user

The 0.0.0.0:8847 is the address to the server (your port number may vary). To view it, simply paste that address to your browser.

The server may take a while to start, depending on if it runs a setup script that could take a bit of time. Files used for the tutorials should have been deployed once the server is ready. If not (and you see empty file list on the left panel of the JupyterLab interface), users should be instructed to manually transfer files to proper folders for use with JupyterLab. More details are documented here.

Setup a cloud VM

Say from a VPS service provider (eg, vultr.com) we purchase a Debian based VM droplet (Debian 9 is what I use as I document this). In the root terminal of the VM,

curl -fsSL https://raw.githubusercontent.com/statgenetics/statgen-courses/master/src/vm-setup.sh -o /tmp/vm-setup.sh
bash /tmp/vm-setup.sh

Create user accounts on cloud VM

I provide a shortcut to create new users on the cloud:

statgen-setup useradd --my-name student --num-users 10

It will generate a password for the user, add it and print the new user ID and password.

Shutdown all running containers

For maintainance, to shutdown all containers and clean up the dangling ones,

statgen-setup clean

This command is only available to root user. For adminstrators I suggest you run statgen-setup clean from time to time to maintain the server. This will terminate running past tutorials in order to free up resources for new tutorials. Otherwise with too many tutorial containers running the same time on a VM it may run out of memory.

Render HTML static website for notebooks

That is, generate https://statgenetics.github.io/statgen-courses/notebooks.html. To do this you need to have sos installed on your local computer if you don't already:

pip install sos -U

To generate the website,

./release

To publish the website, simply add contents in docs/ folder to the github repository and push to github.

Convert (roughly) from MS Word to Notebook

Commands below will provide a rough conversion from docx file (doc files will not work!) to Notebook file:

pandoc -s exercise.docx -t markdown -o exercise.md
notedown exercise.md > exercise.ipynb

The notedown program can be installed via pip install notedown. Additionally you need to install pandoc program.

The conversion is just a start point. Manual polishment is still needed after the automatic conversion. Specifically, it will be important to separate codes from text to different Notebook cells, and assign to each cell the approperate kernel if using SoS multi-language Notebook. Command output should also be removed from the text because they will be generated automatically and formatted better, after executing the notebook.