RStudio

A template project for challenge analysis in R and Python

Introduction

The motivation for this project is to encourage the use of portable development environments in research and engineering. The environment should be intuitive to use so that anyone can deploy it and reproduce your results - even you six months from now!

This project provides a portable development environment that enables you to run the analysis included in this repository (see Notebooks). The Docker image sagebionetworks/challenge-analysis provided by this project and that enables to run the notebooks seamlessly is based on the image sagebionetworks/rstudio.

For more information on how to use this repository to develop and publish your own analysis, please read the section How to use this repository.

All packages:

• R (see renv.lock).
• Python (see conda/sage-bionetworks/environment.yml).

Requirements

• Docker Engine >=19.03.0

Notebooks

The notebooks below are rendered to HTML and published to GitHub Pages by the CI/CD workflow of this repository.

Rmd Notebook	Description	HTML Notebook
compare-models-to-baseline.Rmd	A simple description of a bootstrap analysis to determine the performance of participants relative to a comparator model.
determine-top-performers.Rmd	A simple description of a bootstrap analysis to determine the top performers in a challenge.
ensemble-analysis.Rmd	A simple description of an ensemble analysis for a challenge.
survey-analysis.Rmd	A simple description of a post-challenge survey analysis.

Important: Please make sure when you write your own notebooks that no sensitive information ends up being publicly available. Please check with the information security officer of your organization to confirm that the approach described here can be applied to your use case.

Usage

1. Create and edit the configuration file. You can initially start RStudio using this configuration as-is.

   cp .env.example .env
   

2. Start RStudio. Add the option -d or --detach to run in the background.

   docker compose up
   

RStudio is now available at http://localhost. On the login page, enter the default username (rstudio) and the password specified in .env.

To stop RStudio, enter Ctrl+C followed by docker compose down. If running in detached mode, you will only need to enter docker compose down.

How to use this repository

You can use the image sagebionetworks/challenge-analysis as-is to start an instance of RStudio and develop tools that interact with Sage Bionetworks services, e.g. Synapse.

If you want to create a portable development environment, start by creating a new GitHub repository from this template. You can then customize your environment by specifying the R and Python packages to include with your image. Finally, edit the the GitHub workflow .github/workflows/ci.yml to indicates the Docker repository where the image should be pushed (see Section Versioning).

Example projects that use this repository / image:

• TBA

Manage R and Python dependencies

R

In RStudio, use the following options to add and update libraries:

• Tools > Install Packages...
• Tools > Check for Package Updates...

Run the command renv::snapshot() to update the file renv.lock, which is used in Dockerfile to install the required R libraries.

Python

See the content of the folder conda for an example of how to define a conda environment. The packages to add to this environment must be added to the file requirements.txt. The creation of one or more Conda environments can be specified in Dockerfile.

Setting Synapse credentials

Set the environment variables SYNAPSE_TOKEN to the value of one of your Synapse Personal Access Tokens. If this variable is set, it will be used to create the configuration file ~/.synapseConfig when the container starts.

Using Conda

This Docker image comes with Miniconda installed (see below) and an example Conda environment named challenge-analysis. This environment includes packages used to interact with the collaborative platform Synapse developed by Sage Bionetworks.

From the terminal

Attach to the RStudio container (here assuming that challenge-analysis is the name of the container). For better safety, it is recommended to work as a non-root user. You can then list the environments available, activate an existing environment or create a new one.

$ docker exec -it challenge-analysis bash
container # su yourusername
container $ conda env list
container $ conda activate challenge-analysis

From RStudio

The R code below lists the environment available before activating the existing environment named challenge-analysis.

> library(reticulate)
> conda_list()
    name                              python
1 miniconda           /opt/miniconda/bin/python
2      challenge-analysis /opt/miniconda/envs/challenge-analysis/bin/python
> use_condaenv("challenge-analysis", required = TRUE)

Setting user / group identifiers

When using Docker volumes, permissions issues can arise between the host OS and the container. You can avoid these issues by letting RStudio know the User ID (UID) and Group ID (GID) it should use when creating and editting files so that these IDs match yours, which you can get using the command id:

$ id
uid=1000(kelsey) gid=1000(kelsey) groups=1000(kelsey)

In this example, we would set USERID=1000 and GROUPID=1000.

Giving the user root permissions

Set the environment variable ROOT=TRUE (default is FALSE).

Accessing logs

docker logs --follow challenge-analysis

Generating an HTML notebook

This Docker image provides the command render that generates an HTML or PDF notebook from an R notebook (.Rmd). Run the command below from the host to mount the directory $(pwd)/notebooks where the R notebook is and generate the HTML notebook that will be saved to the same directory with the extension .nb.html.

docker run --rm \
    --env-file .env \
    -v $(pwd)/notebooks:/notebooks \
    sagebionetworks/challenge-analysis:latest \
    render /notebooks/*.Rmd

Versioning

GitHub tags

This repository uses semantic versioning to track the releases of this project. This repository uses "non-moving" GitHub tags, that is, a tag will always point to the same git commit once it has been created.

Docker tags

The artifact published by this repository is the Docker image sagebionetworks/challenge-analysis. The versions of the image are aligned with the versions of R/RStudio, not the GitHub tags of this repository.

The table below describes the image tags available.

Tag name	Moving	Description
latest	Yes	Latest stable release.
edge	Yes	Lastest commit made to the default branch.
weekly	Yes	Weekly release from the default branch.
<major>	Yes	Latest stable major release of this analysis.
<major>.<minor>	Yes	Latest stable minor release of this analysis.
<major>.<minor>.<patch>	Yes	Latest stable patch release of this analysis.
<major>.<minor>.<patch>-<sha>	No	Same as above but with the reference to the git commit.

You should avoid using a moving tag like latest when deploying containers in production, because this makes it hard to track which version of the image is running and hard to roll back.

License

Apache License 2.0