BUG: module load `R/4.1.2` resolves to R 4.3.3 on `bioshell_test`

[fredjaya]

Summary

Loading the R/4.1.2 module on the bioshell_test image sets PATH to /apps/R/4.1.2/bin/R as expected, but the binary at that path reports itself as R 4.3.3. This causes downstream failures when using Bioconductor, which selects its release version based on the running R version.

Steps to reproduce

R module is available:

module avail R

------------------------------------------------ /apps/Modules/modulefiles ---------------------------------------------
   R/4.1.2           jupyter/2024.10     nf-core/3.5.2        snakemake/6.15.1
   ansible/2.16.3    nextflow/25.10.4    rstudio/2023.12.1
------------------------------------------------ /opt/Modules/modulefiles ----------------------------------------------
   shpc    singularity
...

module load R/4.1.2
R --version

Expected behaviour:

R version 4.1.2 (2021-11-01)

Actual behaviour:

R version 4.3.3 (2024-02-29)

The module loads without error and which R correctly points to /apps/R/4.1.2/bin/R, suggesting the binary at that path is not actually R 4.1.2.

Other testing

Check that it is calling the module wrapper from /apps:

which R

/apps/R/4.1.2/bin/R

Unload the module and check R again:

module purge
which R

/usr/bin/R

module spider R

-----------------------------------------------------------------------------------------------------------------------
  R: R/4.1.2
-----------------------------------------------------------------------------------------------------------------------
     Other possible modules matches:
        jupyter, nf-core, rstudio, singularity
    This module can be loaded directly: module load R/4.1.2
-----------------------------------------------------------------------------------------------------------------------
  To find other possible module matches execute:
      $ module -r spider '.*R.*'

Attempting to install Bioconductor packages after loading R/4.1.2 fails with:

Error: Bioconductor version '3.14' requires R version '4.1'; use `version = '3.18'`
  with R version 4.3; see https://bioconductor.org/install

Bioconductor 3.14 is the correct release for R 4.1.x. Because the running binary is actually 4.3.3, the version check fails and installation cannot proceed as configured.

[eileen-xue]

R version mismatch is fixed in the new updates, shown as R/4.3.3.