askrajiv/carte-r repository overview

⁠carte-r

⁠Containerized Analysis Run Time Environment for R

⁠Overview

carte-r provides a customizable Docker container for data analysis using the R programming language.

• Github source: askrajiv/carte-r⁠
• Alternate image with a static RStudio baked in (faster startup, no runtime installation): askrajiv/carte-r-rstudio⁠

⁠Aims

This project grew out of a need to run various scientific-computing and data-science workflows in R across multiple hosts in a reproducible manner. In particular the requirements were:

• support for a configurable and feature-rich computational environment frequently needed to process and explore real-world data
• an IDE for exploratory data analysis and developing and testing R code
• the ability to quickly expose custom algorithms via APIs for enabling web applications

⁠Features

carte-r is a single Docker image that provides command-line access to a customized R environment with the ability to enable additional software components (like rstudio, opencpu and shiny) at runtime.

The image integrates several software components developed by others:

• The R analysis environment is based on the r-apt image from the rocker project⁠

  • The image includes a version specific R installation with 200+ packages from CRAN and Bioconductor focused primarily on computational biology, data analysis and report generation.

  • In addition pre-configured CRAN and the marutter PPA repositories provide easy access to to over 4000+ pre-compiled binary versions of most frequently employed R packages

• Several additional services that can be optionally enabled at runtime based on :

  • Browser based IDE for R development via Rstudio Server⁠
  • OpenCPU⁠ server to create HTTP based API
  • Shiny⁠ server to serve R-based interactive web applications

• The s6-overlay⁠ init system to manage multiple processes in a single container

⁠Getting Started

⁠Using the docker command

• Start a container with just R + packages and mount the host $HOME folder at /work within the container

docker run -v $HOME:/work -it -e 'PASSWORD=changeme:)' -e 'USERID=$UID' --rm quay.io/rajivnarayan/carte-r:latest /bin/bash

• Start r-studio server

docker run -p 8787:8787 -p 3838:3838 -p 8080:80 -v $HOME:/work -it -e 'PASSWORD=letmein:)' -e ADD_RSTUDIO=true -e SUDO=true -e USERID=$UID --rm quay.io/rajivnarayan/carte-r:latest /bin/bash

⁠Using docker-compose

• Initial setup

  • Clone this repo and change

  cd carte/carte-r

  • Copy the dotenv file to .env and change the variables as needed

  cp dotenv to .env

• Start a base container with command-line access

docker-compose run base

Type R to start the R interpreter

• Start a full-fledged container with Rstudio and opencpu

docker-compose up full

Access Rstudio by visting http://localhost:8787⁠ in a web browser (change the port to $RSTUDIO_HOST_PORT if changed in the .env file)

Access OpenCpu http://localhost:8080/ocpu⁠ or the value of $OPENCPU_HTTP_PORT in the .env file

⁠Building a new image using the Docker file

docker build --no-cache=true --build-arg BUILD_DATE=$(date -u +'%Y-%m-%dT%H:%M:%SZ') --build-arg BUILD_VERSION="test" -t carte-r:latest

⁠Docker Environment variables

• PASSWORD: Required, sets the carte user password at runtime
• SUDO: Adds carte user to sudoers file if true, Default is false
• USERID: Sets the UID of the user, default is 1000
• GROUPID: Sets the GID, default is 1000
• ADD_RSTUDIO: Install R-Studio Server within the container if true. Default is false
• ADD_OPENCPU: Install OpenCPU if true. Default is false
• ADD_SHINY: Install Shiny serever if true. Default is false

⁠Container ports

Services if enabled run on the following ports within the container:

• RSTUDIO_PORT=8787
• SHINY_PORT=3838
• OPENCPU_HTTP_PORT=80
• OPENCPU_HTTPS_PORT=443