Dear Team,
I need a little support with regards to creating my own custom function and running it on the server side.
I was following the tutorial provided here on " How to upload a server-side function" (D2K Wiki)
After creating a simple function following the instructions, i have no idea on how to call this function in RStudio, for instance.
Can anyone please provide or point me to a tutorial on how to do this?
Thank you in advance.
Hi @ayisharuna
You need to create a client-side function that will call your server-side function. The call is done through the DSI function datashield.aggregate() if your server-side function is an aggregate function or datashield.assign() if your server-side is an assign one.
I suggest you to have a look on an existing pair of client/server functions that you can find on DataSHIELD repositories on github and see how the datashield.aggregate() and datashield.assign() functions are used from a dsBaseClient function. Probably a simple function that you can have a look is the pair ds.abs()/absDS() (this is an assign function) or the pair ds.levels()/levelsDS() (this is an aggregate function).
Please have a look and we can discuss further if you need any help.
Many thanks, Demetris
Hello @ayisharuna ,
When I finally understood the esential basics of DataSHIELD development I wrote a very concise guide to help new developers understand easily what’s going on. You can find it here:
You will see that I focused on local development using the DSLite package, which emulates a server. If you are further interested or this document did not answer your questions let me know.
Cheers
@demetris.avraam thank you very much for the information and for your time. I will take a look as you suggested.
Best regards, Ayisha.
@xescriba This is very useful. Thank you so much. I will have a look at the link you provided and I will come back if I have any further questions.
Best regards, Ayisha.
Can you please also test your function using the testing framework.
https://data2knowledge.atlassian.net/wiki/spaces/DSDEV/pages/658505761/Testing
@alexwesterberg can we link to this developer guide for DataSHIELD on the DataSHIELD wiki. I’m not sure if the developer tutorial was ever finished.
@xescriba So I finally got down to follow the instructions in the documentation you provided; thank you very much, it was very helpful. I tested my function with DSLite as suggested and I got the result I wanted.
However, I have my own Opal installation on a server, I am wondering how I can utilize the line below to include dsBase and dsX for my own server:
dslite.server ← newDSLiteServer(config = DSLite::defaultDSConfiguration(include=c(“dsBase”, “dsX”)))
Do i include it when I build the login dataframe? A short example will be very much appreciated.
Thank you in advance
@ayisharuna I’m really glad the documentation was helpful.
If you have your own Opal server, one option would be to install your dsX package using the UI (11 Setup | Orchestrating privacy-protected big data analyses of data from different resources with R and DataSHIELD).
Another option is to use opalr to comunicate with the Opal server using an R console. The following code for example:
library(opalr)
o <- opal.login(username = "user", password = "pass", url = "https://your-opal-URL.org")
opal.execute(o, script = 'devtools::install_github("user/dsX")', async = TRUE)
dsadmin.set_package_methods(o, "dsX")
opal.logout(o)
I hope this is solves your doubt.
Hi,
There is an even better way to install a DataSHIELD package in Opal: with dsadmin.install_local_package no need to push your local changes in github.
o <- opal.login('administrator','password', url='https://opal-demo.obiba.org')
# build archive file from local package source (in current working folder)
dsadmin.install_local_package(o, devtools::build())
opal.logout(o)
See also the other dsadmin.* functions for managing packages.
Regards
Yannick
Thank you for the information @xescriba and @yannick . I had started with the option to first push my local changes to github but ultimately I resolved for the option by @yannick that did not require this. Still it was reading your documentation @xescriba that prompted me to make sure I set the right working directory in RStudio before running “dsadmin.install_local_package(o, devtools::build())”. So thank you both for your time!
Hey, I followed your instructions and also fantastic documentation page 21 Tips and tricks | Orchestrating privacy-protected big data analyses of data from different resources with R and DataSHIELD. I am interested in adding packages through this command
dsadmin.install_local_package(o, path=packageArchivePath)
and everything works, but package is stored in /var/lib/rock/R/library on rock while I install manually from UI package lands in /usr/local/lib/R/site-library. From my perspective this is uncomfortable, because we restart docker container every day, therefore data from /var/lib/rock/R... are lost. Persistent storage is mounted to /srv directory. Ive tried to change “LibPath” of installed packages by setting env var (ENV ROCK_LIB /var/lib/rock/R/library), but it doesnt work. So, the question is how I can change “LibPath” to another location like to /srv. Any other ideas are welcome
Hi,
You can probably fine tune the rock container’s default R library folder location by adding this line to the mounted file /srv/conf/Rprofile.R:
.libPaths(append("/srv/R/library", .libPaths()))
Regards,
Yannick