Testbed for R/DataShield with minimal local installation

Hallo DS community,

I wonder whether there is some testbed available to apply R/DataSHIELD without going through the trouble of setting up a Federated Analyses infrastructure with multiple data host servers. I would like to just have to install R/DataSHIELD on my local laptop and connect it with some other servers to do some test runs, or maybe just through a webinterface?

Best, Wilmar

Hi,

You can have a look at DSLite (available in the CRAN) which is a server-less implementation of the DataSHIELD interface. Be aware that the behavior of DSLite is slightly different: the R function calls that can be sent to the “R servers” is not strictly verified, as Opal would do. But it is still a great tool to get familiar with the concept of federated analysis.

Regards
Yannick

1 Like

We are also working on a sandbox server that you could test against from your local R client. I hope this will be ready in a few days

1 Like

Hi Wilmar,

Yannick’s suggestion is a good one. As an alternative to it, the DataSHIELD training material does just set up a VM locally on your computer to act as a federated analysis infrastructure, without really having to set one up. Instructions are found at the Installation and Training Hub, here: https://data2knowledge.atlassian.net/wiki/spaces/DSDEV/pages/931069953

But if this is something you’ve already looked at and decided against, that’s quite alright!

@Yannick would it be possible to use opal-demo in the short term as another option?

@tombishop yes… as soon as people are making a reasonable usage of it :slight_smile:

@wilmar.igl you can use https://opal-demo.obiba.org with credentials administrator/password, system is rebuilt daily. There are some “DataSHIELD” projects with simulated test data (CNSIM, DASIM, etc.) that you can use. But I still recommend to have a look at DSLite, so that you could play with your own datasets.

Yannick

Thanks for many great ideas to check out! Happy to see that the DS community is so active!

1 Like