Run the Cobrapy stack in docker

This repo helps you to build a docker container running cobrapy on Python 3 and a jupyter notebook server apart from those two it also provides the following "add-ons":

• optional https and passwords protection
• reconstruction of metabolic models using CORDA
• unpriviliged user
• Latex to export notebooks as PDF
• matplotlib and seaborn for plotting
• installation of commercial solvers (only cplex, please contact me or send a pull request if you want to help out with gurobi and mosek)

Installation

The image is built automatically by the docker automated build system. To get the latest version just pull it in docker. Obviously, this will NOT include commercial solvers.

docker pull cdiener/cobra-docker

Building locally

A local version including commercial solvers and customized models can be built with

mkdir solvers
docker build -t cobra .

Commercial solvers

Cplex

In order to have cplex installed you will need the current cplex install (12.6+) and extract its contents to the "solvers" directory into a subfolder "ibm". You can do that by running the Cplex Installer and when prompted for the installation directory choosing "solvers/ibm" in the directory from where you will build your docker container. The Dockerfile will take care of the rest.

Environment variables

This container supports some of the environment variables of the minimal Jupyter image. The most important ones are:

• using the the start-notebook.sh script to directly write to the jupyter config
• GEN_CERT - to use https (will require you to access the notebook via https://localhost:PORT)

So you can run a notebook server with HTTPS and password "notebook" at https://localhost:8888 with

docker run -p 8888:8888 -e GEN_CERT=yes cobra start-notebook.sh --NotebookApp.token='notebook'

Planned changes

• [X] unprivileged user
• [ ] more solvers
• [X] pdf export?