FATAL ERROR after mamba installation

[jzrapp]

Hi,

I'm trying to run dadaist2 on macOS 14.5, but unfortunately I am unable to replicate the tutorial.
I installed dadaist2 using:
mamba create -n dadaist -c conda-forge -c bioconda dadaist2
When I run dadaist2 -i data/16S/ -o example-output -d refs/SILVA_SSU_r138_2019.RData -t 2 -m metadata.tsv --verbose, I receive this output:

`[2024-07-02 16:54:14] Ready to log in /Users/jzrapp/example-output/dadaist.log
[2024-07-02 16:54:14] dadaist2 1.3.0
[2024-07-02 16:54:14] DECIPHER Taxonomy database found: refs/SILVA_SSU_r138_2019.RData
Use of uninitialized value in concatenation (.) or string at /Users/jzrapp/miniconda/envs/dadaist2/bin/dadaist2 line 535.
[2024-07-02 16:54:14] Parameter: taxonomy-type:
[2024-07-02 16:54:14] Parameter: taxonomy-db: skip

• Input directory: data/16S/
• Output directory: example-output/
• Metadata: metadata.tsv
• Reference database: skip
• Threads: 2
• Temporary directory: /var/folders/_m/_2ckc70s5q776c448cvkx2n40000gn/T/dadaist2_mswkKi
• QC strategy: skip
  [2024-07-02 16:54:14] QC: Checking quality profile with SeqFu
  [2024-07-02 16:54:15] SeqFu quality truncation at (trunc-len-1 and trunc-len-2): 277 - 224
  [2024-07-02 16:54:15] Checking dependencies
  FATAL ERROR:
  Dependency 'DECIPHER' returned non zero exit value.
  Shell: echo "library(DECIPHER)" | R --no-save > /dev/null
  Dadaist2 execution finished (4.00s)`

I thought there was an issue with DECIPHER, so I also tried running it with a different database dadaist2 -i data/16S/ -o example-output -d refs/silva_nr_v138_train_set.fa.gz -t 2 -m metadata.tsv --verbose

`
[2024-07-02 16:54:28] Ready to log in /Users/jzrapp/example-output/dadaist.log
[2024-07-02 16:54:28] dadaist2 1.3.0
[2024-07-02 16:54:28] Taxonomy database found: refs/silva_nr_v138_train_set.fa.gz
[2024-07-02 16:54:28] Parameter: taxonomy-type: dada2
[2024-07-02 16:54:28] Parameter: taxonomy-db: refs/silva_nr_v138_train_set.fa.gz

• Input directory: data/16S/
• Output directory: example-output/
• Metadata: metadata.tsv
• Reference database: refs/silva_nr_v138_train_set.fa.gz
• Threads: 2
• Temporary directory: /var/folders/_m/_2ckc70s5q776c448cvkx2n40000gn/T/dadaist2_2EmI4D
• QC strategy: skip
  [2024-07-02 16:54:28] QC: Checking quality profile with SeqFu
  [2024-07-02 16:54:28] SeqFu quality truncation at (trunc-len-1 and trunc-len-2): 277 - 224
  [2024-07-02 16:54:28] Checking dependencies
• RScript: R scripting front-end version 4.0.5 (2021-03-31)
• Taxonomy: dadaist2-assigntax 1.1.3
• assign-taxonomy: dadaist2-assigntax 1.1.3
• clustalo: 1.2.4
  FATAL ERROR:
  Dependency 'dada2 (lib)' returned non zero exit value.
  Shell: echo "library(dada2)" | R --no-save > /dev/null
  Dadaist2 execution finished (2.00s)
  `

I am not sure what the problem is. I thought maybe some dependencies were not installed correctly, so I also tried to install dadaist2 via mamba env create --file dadaist2-1.2.5-Darwin.yaml -n dadaist2 but it failed and gave a very long list of reasons, which mostly said conflicts with any installable versions previously reported.

Do you have any idea what the issue might be, @telatin?
Thanks a lot!

[jzrapp]

Hi, I think I found the issue and have also found a semi-permanent solution.

The issue arises from R .libPaths() that conflicts with user/system installed R. So even when creating a separate env for the mamba/conda installation, the .libPaths() defaults to my system installation and cannot locate all libraries. The issue is described here in detail:
conda-forge/r-base-feedstock#37
picrust/picrust2#139

For now, I ended up changing my .Rprofile file and added
.libPaths('/Users/jzrapp/miniconda/envs/dadaist/lib/R/library')
and then used
.libPaths( c(.libPaths()[2], .libPaths()[1]) )

to specify the order in which the libraries should be used.
I am now able to run the tutorial without issues, but I think it would be nice to have a more elegant solution... :)

[telatin]

Thanks for reporting! I never had this issue on my setups but I'm not an expert R user.
I'll try to mention this in the tutorial!