Rstudio server in Snakemake

Question

[sakura]

Hello, I am new to using Snakemake and I am running a test script calling an R script. I work on a restricted server and I have access to a docker R studio server to run everything. However, it seems I am not able to connect the Snakemake to the Rstudio docker. Is that possible? Or do I need to install R?

This is my R script:

args = commandArgs(trailingOnly = TRUE)

num_cells = args[[1]]

write.table("Cells: 1000", num_cells)

This is the Snakemake script (Snakemake_Test1_PS.txt):

rule test_file:
    output:
        num_cells="all_cells.txt"
    shell:
        "Rstudio --no_environ Scripts/testing_snakemake.R {output.num_cells}"

Output:

$ snakemake -s Snakemake_Test1_PS.txt
Building DAG of jobs...
Using shell: /usr/bin/bash
Provided cores: 88
Rules claiming more threads will be scaled down.
Job counts:
        count   jobs
        1       read_file
        1

[Tue Jul 18 12:38:20 2023]
rule read_file:
    output: all_cells.txt
    jobid: 0

/usr/bin/bash: Rstudio: command not found
[Tue Jul 18 12:38:20 2023]
Error in rule read_file:
    jobid: 0
    output: all_cells.txt
    shell:
        Rstudio --no_environ Scripts/testing_snakemake.R all_cells.txt
        (one of the commands exited with non-zero exit code; note that snakemake uses bash strict mode!)

Thank you!!

[Asaf]

Why do you run Rstudio and not simply R?

[sakura]

I would like to run it in the R studio server environment I have. But either way just R doesn't work either. I get the exact same error

[Asaf]

So do you run the script inside the docker? Do you run this specific part in the docker?

[sakura]

No, it doesn't work inside the docker because Snakemake isnt recognized. I run the docker using:

sudo docker exec -it rstudio_server4.2 bash

but mostly I use the Rstudio webpage interface linked to the docker

[Asaf]

1. You should avoid running anything with sudo, there is a way to run docker without a sudo, see here.
2. You should tell snakemake in which environment it should run the rule. See here , you should add a docker: label to your rule.

[sakura]

Unfortunately I cannot do that. Although I have sudo access I wasn't given the docker account password. I understand that doesn't really make sense.

Running this:

rule test_file:
    output:
        num_cells="all_cells.txt"
    container:
        "docker://joseespinosa/docker-r-ggplot2"
    script:
        "Scripts/testing_snakemake.R"

Gives me this error

SyntaxError in line 4 of /redacted/Snakemake_Test1_PS.txt:
Unexpected keyword container in rule definition (Snakemake_Test1_PS.txt, line 4)

[Asaf]

What version of snakemake are you using?

[sakura]

It is version 5.9.1

[Asaf]

container support was introduced in 4.8.0 so it should work, but your version is quite old. I recommend updating to the latest version and giving it another try.

[camerond]

How do you normally use R/Rsudio on the server? It would appear that it is not installed / available. On our server, R is not available by default and requires running module load R to make it available. If it is not available at all you will need to install R and Rstudio for these commands to run.

[sakura]

I think thats my issue. I don't know how to have the console access the docker while still using the usual bash interface where Snakemake is. Or if that is even possible

Answer 1

[DBScan]

There are several ways you could do, I would suggest you make use of the container keyword in the test_file rule (see here Snakemake container). Generally it's also a good idea to use script instead of shell if you call R scripts (see here R script).