OHDSI Home | Forums | Wiki | Github

How to deploy the Data Quality Dashboard with ShinyProxy?

General
#1

Hi everyone,

I’m new to R Shiny deployment. Currently, I have executed ‘DataQualityDashboard::executeDqChecks’, generating a JSON file, and I am able to visualize the results using ‘viewDqDashboard(jsonFilePath)’ locally. My next step is to deploy the dashboard to internal server. I am using Windows.

I noticed that typical Shiny apps have ‘ui.R’, ‘server.R’, or ‘app.R’ files. In my folder, I only have an R script and the generated ‘results.json’ file.I came across this link https://github.com/OHDSI/ShinyDeploy/tree/master/DataQualityDashboard which contains an ‘app.R’ and a ‘www’ folder. But I don’t understand these work together with ui.R and server.R.

How can I deploy this dashboard on internal server with ShinyProxy? Where can I find more instruction about the Data Quality dashboard deployment?

Thank you.

#2

Hi @Siyan,

I don’t have much experience deploying apps to shinyproxy, however their documentation can be found (here)[Deploying Apps | ShinyProxy] on setting up applications inside docker containers.

You will likely want to make a few modifications for this to copy the json file produced by dqd

...

# copy the app to the image
RUN mkdir /app
COPY <INSERT PATH TO APP> /app

COPY Rprofile.site /usr/lib/R/etc/

# Overwrite results
COPY result.json /app/www/results.json

EXPOSE 3838

CMD ["R", "-e", "shiny::runApp('/app')"]

As we don’t run shinyproxy I’m afraid it will be hard to be more specific

1 Like
#3

Hi @jpegilbert ,

Thanks for your reply. I’m also considering trying Shiny Server. Please let me know if you have any instructions or experience with it. Thank you.

#4

@Siyan I have experience with posit connect (formerly rsconnect) and that couldn’t be easier so I’d strongly recommend it from a user perspective - deployment of OHDSI shiny apps should be supported going forward and the andboxing/isolation you get from use of renv with the deployed apps is very easy.

I’m yet to deploy DQD in this manner myself, but I know my colleagues at JNJ have and any code on CRAN or github (like all our OHDSI R packages) is easy to install.

t