There are currently two approaches available for deploying Shiny apps, either manually via Shinyapps.io, or via a SciComp supported pipeline. Which approach is best for a given project depends on what the needs of the app are:
- Using Shinyapps.io to deploy Shiny Apps
- It streamlines with RStudio allowing auto-deployment in the interface.
- There are multiple plans, such as the free plan which currently includes 25 hours activity per month, or the ‘standard’ and ‘professional’ plans which include password protection (both) and custom domain/urls (professional only), for a monthly subscription fee.
- The free plan currently limits users to 5 apps per account!
- Deploying shiny apps on Shinyapps.io does not require the use of GitHub or allows you to include multiple applications within a single GitHub repository if you have many apps to manage.
- Provides a web analytics feature showing app activity.
- Using the application deployment pipeline developed by Fred Hutch Scientific Computing
- SciComp’s pipeline has the capability to host applications written in other languages as long as the application itself can be containerized.
- Integrates with data sources within Hutch campus. For example, the application can be connected to a campus based database system (Postgres, mysql, etc…) without exposing it to the entire internet.
- There is a template repository (https://github.com/FredHutch/shiny-app-template) to help users to assemble your own shiny app.
- Once the application is deployed, updates/re-deploying is done by committing changes to GitHub, as the pipeline is based on a continuous integration/continuous delivery (CI/CD) feature.
- Users can pick their own custom site URL in the fredhutch.org domain (example: ‘myshinyapp.fredhutch.org’).
- User can specify if their app is only facing campus within the firewall or being exposed to the entire internet. Authorization feature can also be included upon request.
- This service is completely free to campus users regardless of usage hours or number of apps.
To deploy a Shiny app via the Fred Hutch system, you must first have access to the Fred Hutch institution GitHub. You can find more information about getting set up with GitHub at Fred Hutch here. As an overview, you must first create a GitHub username, and then have
scicomp connect it to the Fred Hutch institution.
Set up your GitHub Repository
There are two ways to set up your repository, and either is suitable. Note: Documentation for using GitHub at the Fred Hutch is being generated here, and it is a good place to start to find the necessary help you may need to use GitHub to set up your Shiny app.
To do a basic set up via the web, go to GitHub and create a new repository in the Fred Hutch institution for your app. Then clone your new repository to your local computer. In the local folder created, create a folder called
app and put your shiny app files (either app.R, or ui.r/server.R) and any other associated data files needed for the app in the
app directory. (Once you are finished and have tested your app locally, either via R or RStudio, push your edits to GitHub. )
Cloning the Template Set Up
The template for your app can be found in this GitHub Repo (accessible after login):
Clone this repository, remove the unnecessary files listed and add your Shiny app files to the
app directory, before pushing your changes to a new repo on GitHub.
Using Command Line Git
git clone https://github.com/FredHutch/shiny-app-template.git <your_app_folder> cd <your_app_folder> ls -al rm -rf .git ls -al git init echo "# <your_app_folder>" > README.md git add . git commit -m 'initial commit'
After the steps above from a terminal, you have achieved these steps:
- Cloned a template
- Created your own repo
- Add a new git control to this repo
Now it’s time to inject your wonderful shiny app to this template. The goal is to put all your app code base to template’s subfolder ‘app’.
Using the GitHub Desktop Application
To keep track of file changes within a local repo, first add this folder to GitHub Desktop Application:
You can create a remote repo by clicking on the ‘Publish repository’ and select the correct branch (in this case ‘master’ is the branch name). You can specify if you want this repo to be private or not. Also, please make sure this repo is under FredHutch as the organization.
Test Your Application Locally
Open the app in an R console.
- Change directory to subfolder ‘app’ under your app root:
cd <your_app_folder> cd app R
- Via R console,
- Then go to a browser to check the url: http://localhost:7777
Insert Your App into the Template
Use your favorite code editor to add your own shiny app content to this template. Here are a few reminders for the shiny apps with a single R script:
- Please split your app.R or ui.R/server.R in the
appdirectory in the local cloned repository
- If you are using the ui.R/server.R structure, make sure you include ‘library(shiny)’ to both ui.R and server.R files, and in server.R, please remove ‘shinyApp(ui, server)’
- Make sure all files (data, etc) needed by the app are also in the
appdirectory or a subdirectory thereof
Push Edits to GitHub
After you added your own content to this repo, you are ready to commit the changes and push the changes to the remote repo.
- Now your GitHub Desktop Application console looks like this:
- Two files with green crosses on the right have been added as new files. Two existing files have been modified with yellow dots on the right. You can further examine and adjust those changes by click the yellow dot or right click:
- Write a commit message and click on ‘commit to master’ if you want to commit change to your master branch:
- Push committed change from local files to remote repo
Viewing Your App
You should be able to find your app within the repo you created.
Deployment Service from SciComp
Hopefully, you can see your app working as expected in the browser. Please keep a good record of all your dependencies. You may have all the necessary R packages installed on your local machine but SciComp deployment platform needs to know those dependencies. You also want to be ready to answer all the questions below before you contact Scientific Computing team. If you are ready with your codebase and have the answers for the below information required by deployment, then go to https://getshiny.fredhutch.org to submit the application deployment request. This request website is only accessible from Campus computers through wired connection or the
Marconi wireless network.
Is your application codebase version-controlled by Git and in a GitHub repository (repo) ? If so, is it under FredHutch account?
How do you want to name (DNS name) your application? So the it will look something like
Is your application facing Fred Hutch internally (open to the Fred Hutch campus) or externally (open to the entire internet)?
Does your application require authorization (password protection), that only your collaborators have access?
Does your application contain PHI? Please remove PHI from your application?
Please list out all the dependencies (R packages you used).