Error, cannot locate genome_lib_dir: ctat_genome_lib_build_dir at /opt/conda/envs/nf-core-rnafusion-star-fusion_1.8.1/lib/STAR-Fusion/STAR-Fusion line 439.

[valleema]

Hello!

First of all, I see all the hard work you are putting in this code, I thank you very much for it (I am following the dev branch closely).

I was able to download all that is required and I was trying my hand on the main.nf script.

I tried it with a test sample which goes well under last version of the pipeline (version as without Arriba - old).

However, star_fusion process ended up with an error (in the title):

Error executing process > 'star_fusion (RNA053-T_S24)'

Caused by:
  Process `star_fusion (RNA053-T_S24)` terminated with an error exit status (2)

Command executed:

  STAR \
      --genomeDir star-index \
      --readFilesIn RNA053-T_S24_R1.fastq.gz RNA053-T_S24_R2.fastq.gz \
      --twopassMode Basic \
      --outReadsUnmapped None \
      --chimSegmentMin 12 \
      --chimJunctionOverhangMin 12 \
      --alignSJDBoverhangMin 10 \
      --alignMatesGapMax 100000 \
      --alignIntronMax 100000 \
      --chimSegmentReadGapMax 3 \
      --alignSJstitchMismatchNmax 5 -1 5 5 \
      --runThreadN 12 \
      --outSAMstrandField intronMotif  \
      --outSAMunmapped Within \
      --outSAMtype BAM Unsorted \
      --outSAMattrRGline ID:GRPundef \
      --chimMultimapScoreRange 10 \
      --chimMultimapNmax 10 \
      --chimNonchimScoreDropMin 10 \
      --peOverlapNbasesMin 12 \
      --peOverlapMMp 0.1 \
      --readFilesCommand zcat \
      --sjdbOverhang 149 \
      --chimOutJunctionFormat 1
  
  STAR-Fusion \
      --genome_lib_dir ctat_genome_lib_build_dir \
      -J Chimeric.out.junction \
      --left_fq RNA053-T_S24_R1.fastq.gz --right_fq RNA053-T_S24_R2.fastq.gz \
      --CPU 12 \
      --examine_coding_effect \
      --output_dir . 
  
  mv star-fusion.fusion_predictions.tsv RNA053-T_S24_star-fusion.tsv
  mv star-fusion.fusion_predictions.abridged.tsv RNA053-T_S24_abridged.tsv
  mv star-fusion.fusion_predictions.abridged.coding_effect.tsv RNA053-T_S24_abridged.coding_effect.tsv

Command exit status:
  2

Command output:
  Jun 15 13:47:07 ..... started STAR run
  Jun 15 13:47:07 ..... loading genome
  Jun 15 13:47:39 ..... started 1st pass mapping
  Jun 15 13:49:24 ..... finished 1st pass mapping
  Jun 15 13:49:25 ..... inserting junctions into the genome indices
  Jun 15 13:50:50 ..... started mapping
  Jun 15 13:53:11 ..... finished mapping
  Jun 15 13:53:12 ..... finished successfully

Command error:
  Error, cannot locate genome_lib_dir: ctat_genome_lib_build_dir at /opt/conda/envs/nf-core-rnafusion-star-fusion_1.8.1/lib/STAR-Fusion/STAR-Fusion line 439.

Tell me if you need more details.

[matq007]

Seems like you are missing ctat_genome_lib_build_dir folder. It should be located in you reference folder.

[valleema]

Mmh yes I do have it. But it has been extracted as a tar should be, keeping the file structure:

data1/references/rnafusion/star_fusion/1.8.1/ctat_genome_lib_build_dir/*

Therefore, in my references folder I have this structure (publishDir in copy mode kept data1/references/rnafusion/star_fusion/1.8.1/ preffix):

rnafusion/references/star-fusion/data1/references/rnafusion/star_fusion/1.8.1/ctat_genome_lib_build_dir/

[matq007]

Hey @valleema, we have updated a lot the configurations as well. Here is a link how it looks: https://github.com/nf-core/rnafusion/blob/dev/conf/genomes.config. When you download references you should already have the correct structure. Please check that you follow this directory path if you have a custom one.

[valleema]

Yes I can see the config is correct at this line:

star_fusion_ref     = "${params.genomes_base}/star-fusion/ctat_genome_lib_build_dir"

Personally, I did not create custom folder structure, I let dowload_references.nf do its job. At this point:

tar -xf ctat_star_fusion_1_8_1.tar.gz

It extracts keeping the folder structure, starting by data1/references/rnafusion/star_fusion/1.8.1/ctat_genome_lib_build_dir/...

However, the process download_star_fusion is in copy mode to ${params.outdir}/star-fusion at this line.

So, the data ends up actually in ${params.outdir}/star-fusion/data1/references/rnafusion/star_fusion/1.8.1/ctat_genome_lib_build_dir/ and main.nf creates an empty symlink to the expected folder ${params.outdir}/star-fusion/references/star-fusion/ctat_genome_lib_build_dir/.

There is an extra data1/references/rnafusion/star_fusion in the path.

[valleema]

As a test, I just created a symlink at the base of the folder:

$ ls -lG
total 1
lrwxrwxrwx 1 valleemax 70 Jun 16 08:34 ctat_genome_lib_build_dir -> data1/references/rnafusion/star_fusion/1.8.1/ctat_genome_lib_build_dir
drwxr-xr-x 3 valleemax  1 Jun 12 10:45 data1

and the symlink in the work folder is no longer broken. Maybe the tar extraction in the process download_star_fusion should be followed by a mv to get rid of the extra arborescence.

[matq007]

Thank for thorough analysis 👍 . I think I have an idea how to fix this.

[matq007]

Ok I see the problem. The command should be tar -xf ctat_star_fusion_1_8_1.tar.gz --strip-components=5. I will update the change

[matq007]

Fixed in ccde848.

[valleema]

Thanks!