Opal R Session

Yannick Marcon

2021-08-23

Opal is connected with one or more R servers. Each Opal user having the appropriate permissions can create a R session on the server side through Opal. See the R Server documentation page that explains the relationship beteen Opal and the R servers. The Opal R package exposes data management related functions:

Note that these functions do create a R session on the server side.

Setup

Setup the connection with Opal, specifying that the R server session must be created with the "default" profile, i.e. in the "default" R servers cluster (note that the profile parameter is optional):

library(opalr)
o <- opal.login("administrator", "password", "https://opal-demo.obiba.org", profile = "default")

Assign

Assign a Opal table to a data.frame (if user has permission to see values of the Opal table):

opal.assign.table(o, symbol = "df", value = "CNSIM.CNSIM1")

In place of a standard data.frame, Opal can assign data into a tibble:

opal.assign.table.tibble(o, symbol = "tbl", value = "CNSIM.CNSIM1")

Resource objects can be assigned to the R server session:

opal.assign.resource(o, symbol = "rsrc", value = "RSRC.CNSIM3")

An R expression can also be assigned to a symbol (use quote() to not interpret the expression in the client side R session). As an example assign a function body:

opal.assign.script(o, symbol = "hello", value = quote(function(x) { paste0("Hello ", x, "!") }))

List the R symbols that lives in the remote R session:

opal.symbols(o)

Execute

R expressions can be executed on the server side:

# column names
opal.execute(o, script = "names(tbl)")
# get variable description from column attributes
opal.execute(o, script = "attributes(tbl$GENDER)")
# coerce resource to a data.frame and plot histogram of one column
plot(opal.execute(o, script = "hist(as.data.frame(rsrc)$LAB_HDL)"))
# execute the custom function
opal.execute(o, script = "hello('friends')")

The remote data can be downloaded into the client side R session:

GENDER <- opal.execute(o, script = "tbl$GENDER")

To execute more complex R code, use a local source file:

opal.execute.source(o, path = "/path/to/some_code.R")

Files

Files can be written from the Opal file system to the R server session folder:

opal.file_write(o, source = "/projects/CNSIM/CNSIM3.zip", "test.zip")

Similarly, files can be read from the R server session folder into the Opal file system:

opal.file_read(o, source = "test.zip", destination = "/tmp")

Workspace

The image of the remote R session can be saved in a workspace (to be restored at login time):

opal.workspace_save(o, save="demo")
opal.workspaces(o)

Teardown

Good practice is to free server resources by sending a logout request:

opal.logout(o)