Problems with loading igenomes on default

[relaxingbob]

bwilson@genomicscore:~$ nextflow run /home/bwilson/rnafusion --star-fusion --genomeDir GRCh38 -profile docker --reads 'data/*_R{1,2}.fastq.gz'

N E X T F L O W ~ version 19.01.0
Launching /home/bwilson/rnafusion-master/main.nf [serene_dubinsky] - revision: ea16375b17
WARN: There's no process matching config selector: download_star_fusion_ensembl
WARN: Access to undefined parameter genomes -- Initialise it to a default value eg. params.genomes = some_value
Missing fromPath parameter

This is running on a workstation with Ubuntu 18.04 with 24 cores and 48G memory.

I tried configuring the igenomes.config file and could get it to find the genome files.

[apeltzer]

Same here...

[matq007]

Ah, I think I know where the problem is: set the parameter igenomes_base in the config file. That should solve it.

[relaxingbob]

In the igenomes.config file, I remove igenomes_base and gave it the complete paths to the genomes file and that didn’t work.

[matq007]

You also have to set genomes to some value like GRCh38.

[relaxingbob]

This is the command line that I used: nextflow run nf-core/rnafusion --star_fusion --genomeDir GRCh38 -profile docker --reads '/data/*_R{1,2}.fastq.gz'

[matq007]

[EDIT]:
So that means you have to tun the pipeline like this instead: nextflow run nf-core/rnafusion --star_fusion --genome GRCh38 --igenomesIgnore false -profile docker --reads '/data/*_R{1,2}.fastq.gz' and then your igenomes.config should be defined like this if you hard-coded paths:

params {
  // illumina iGenomes reference file paths
  genomes {
    'GRCh38' {
      bed12   = "/PATH/TO/genes.bed"
      fasta   = "/PATH/TO/genome.fa"
      gtf     = "/PATH/TO/genes.gtf"
      star    = "/PATH/TO/STARIndex/"
    }

[relaxingbob]

Yes. You are correct.

[relaxingbob]

I don't know if this additional warning will help narrow this down or not: WARN: There's no process matching config selector: download_star_fusion_ensembl

[relaxingbob]

Sorry for the confusion. With this docker, mproksik/rnafusion, everything works fine.

The failure I am having with with nextflow nf-core/rnafusion.

Thanks

[matq007]

Have you tried the updated comment I've left @relaxingbob?

[relaxingbob]

Sorry. I don't see your comment.

[matq007]

Hey @relaxingbob, did you solve the issue?

[apeltzer]

Hi @matq007 !

I had to do two things in here (showing the result of my git diff on a checked out master branch of rnafusion version 1.0.2 here:

git diff
diff --git a/main.nf b/main.nf
index a6a2302..cd2d53f 100644
--- a/main.nf
+++ b/main.nf
@@ -96,7 +96,7 @@ Channel
     .ifEmpty { exit 1, "GTF annotation file not found: ${params.gtf}" }
     .into { gtf; gtf_squid }

-if (!params.star_index && (!params.fasta && !params..gtf)) {
+if (!params.star_index && (!params.fasta && !params.gtf)) {
     exit 1, "Either specify STAR-INDEX or fasta and gtf file!"
 }

diff --git a/nextflow.config b/nextflow.config
index a5a23c6..68c22fc 100644
--- a/nextflow.config
+++ b/nextflow.config
@@ -47,7 +47,6 @@ params {
   clusterOptions = false
   awsqueue = false
   awsregion = 'eu-west-1'
-  readPaths = null
   debug = false

   // Options: download-references.nf
@@ -58,8 +57,8 @@ params {

   // Shared default variables across different scripts
   download_db = false
-  igenomesIgnore = true
   outdir = './results'
+  igenomesIgnore = false
   tracedir = "${params.outdir}/pipeline_info"

   // Boilerplate options

Using this, I was able to use my bash script (as mentioned on slack, @maxulysse was also helping out there - thank you a lot!) to get things in a "batch style mode":

#!/usr/bin/bash
module purge
module load devel/java_jdk/1.8.0u112
module load devel/singularity/3.0.3

##SampleIDs to make Batch mode possible
list=( I15R018a01 I15R018b02 I15R018e02 I16R003a02 I16R003a03 I16R003c01 I16R003d01 I16R003e01 I16R003f01 I16R003g01 I16R003i01 I16R003i02 I16R003j01 I17R018Ra01 I17R018Ra02 I17R018Rc01 I17R018Rd03 I17R018Rf02 I18R020Ra01 I18R020Ra02 I18R020Rg01 I18R020Rh01 )

#Run RNAfusion
for i in "${list[@]}"
do
        echo "Running on sample $i"
    suffix='*{1,2}.fastq.gz'
    prefix='RAW/'
    path=${prefix}${i}${suffix}
    nextflow run ./rnafusion/main.nf --reads "$path" -profile binac --databases '/beegfs/work/iiipe01/2019-05-27_SFB209_RNAfusion/dbs_for_fusion/dbs_for_fusion' --fusioncatcher_ref '/beegfs/work/iiipe01/2019-05-27_SFB209_RNAfusion/dbs_for_fusion/dbs_for_fusion/fusioncatcher_ref/human_v90/' --star_fusion --star_fusion_ref '/beegfs/work/iiipe01/2019-05-27_SFB209_RNAfusion/dbs_for_fusion/dbs_for_fusion/star_fusion_ref/GRCh38_v27_CTAT_lib_Feb092018/ctat_genome_lib_build_dir' --ericscript --ericscript_ref '/beegfs/work/iiipe01/2019-05-27_SFB209_RNAfusion/dbs_for_fusion/dbs_for_fusion/ericscript_ref/ericscript_db_homosapiens_ensembl84' --pizzly --pizzly_fasta '/beegfs/work/iiipe01/2019-05-27_SFB209_RNAfusion/dbs_for_fusion/dbs_for_fusion/pizzly_ref/Homo_sapiens.GRCh38.cdna.all.fa.gz' --pizzly_gtf '/beegfs/work/iiipe01/2019-05-27_SFB209_RNAfusion/dbs_for_fusion/dbs_for_fusion/pizzly_ref/Homo_sapiens.GRCh38.94.gtf' --genome "GRCh38" --outdir "results_$i" --star_index '/nfsmounts/igenomes/Homo_sapiens/NCBI/GRCh38/Sequence/STARIndex/' --igenomes_base '/nfsmounts/igenomes/' -resume -dump-channels
done

[apeltzer]

So it runs well and ~8 samples ran through during the weekend (cluster is more crowded than I anticipated ;-)). I guess I'll use the resume option to also add reports for the other potential tools such as squid to have more complete reports - but this was merely a test balloon and seems to work just fine 👍 Batch mode in-built would be nicest I guess however ;-)

[matq007]

@apeltzer thanks for figuring this out! You 🚀. I will implement the fix tomorrow in the dev branch. I will probably remove igenomesIgnore and just include the genomes.config on default, to make it less confusing.

[apeltzer]

No problem - I just wanted to share that info back since it took me a while 😓

[apeltzer]

30 samples processed, ~350GB crunched! Thanks a bunch for all the input!

[relaxingbob]

I finally got this to work. Thanks for the help.