add vv raw reads stub, fix configs

Author: torres-alexis

RNAseq/Workflow_Documentation/NF_RCP/workflow_code/conf/by_docker_image.config

@@ -60,7 +60,7 @@ process {
         container = "quay.io/biocontainers/rsem:1.3.3--pl526ha52163a_0"
     }
 
-   withName: 'GET_ACCESSIONS|FETCH_ISA|ISA_TO_RUNSHEET|RUNSHEET_FROM_ISA|GENERATE_MD5SUMS|SOFTWARE_VERSIONS|UPDATE_ISA_TABLES|PARSE_QC_METRICS|REMOVE_RRNA_FEATURECOUNTS|MD5SUM' {
+    withName: 'GET_ACCESSIONS|FETCH_ISA|ISA_TO_RUNSHEET|RUNSHEET_FROM_ISA|GENERATE_MD5SUMS|SOFTWARE_VERSIONS|UPDATE_ISA_TABLES|PARSE_QC_METRICS|REMOVE_RRNA_FEATURECOUNTS|MD5SUM' {
         container = "quay.io/nasa_genelab/dp_tools:1.3.5"
     }
 

RNAseq/Workflow_Documentation/NF_RCP/workflow_code/conf/slurm.config

@@ -87,7 +87,7 @@ process {
         cpus   = { 12    * task.attempt }
         memory = { 72.GB * task.attempt }
     }
-    withName:DESEQ2_DGE {
+    withName:DGE_DESEQ2 {
         // To do: Optimize both settings, large dataset wald
         cpus   = { 4     * task.attempt }
         memory = { 16.GB * task.attempt }

RNAseq/Workflow_Documentation/NF_RCP/workflow_code/modules/vv.nf

@@ -1,80 +1,45 @@
-/* VV check processes
-* These processes intentional deviate from Nextflow isolation to ensure QC reports
-*   are based on files in publish directory and not work directories.
-*/
-
-// NOTE: first VV step also creates initial VV file that is shared across all vv steps
-// process VV_RAW_READS {
-//     label 'VV'
-//     // Log publishing
-//     publishDir "${ publishdir }",
-//         pattern: "test.txt",
-//         mode: params.publish_dir_mode
-    
-//     input:
-//         val(publishdir)
-//         val(meta)
-//         path("Metadata/*_runsheet.csv")     // Runsheet
-//         path("00-RawData/Fastq/*")          // Raw reads
-//         path("00-RawData/FastQC_Reports/*") // Raw FastQC reports 
-//         path("00-RawData/FastQC_Reports/*") // Raw MultiQC report
-//         path(dp_tools__NF_RCP)
-
-//     output:
-//         path("test.txt"), emit: test
-
-//     script:
-//         """
-//         echo "Meta: ${meta}" > test.txt
-//         """
-// }
 process VV_RAW_READS {
-  // Log publishing
+  // Publish VV log
   publishDir "${ publishdir }",
     pattern:  "VV_log.tsv" ,
     mode: params.publish_dir_mode,
     saveAs: { "VV_Logs/VV_log_${ task.process.replace(":","-") }${ params.assay_suffix }.tsv" }
-  // V&V'ed data publishing
+  // Publish VVed data
   publishDir "${ publishdir }",
     pattern: '00-RawData/**',
     mode: params.publish_dir_mode
 
   label 'VV'
 
   input:
+    path(dp_tools__NF_RCP)
     val(publishdir)
     val(meta)
-    path("VV_INPUT/Metadata/*")                  // Runsheet
-    path("VV_INPUT/00-RawData/Fastq/*")          // Raw reads
-    path("VV_INPUT/00-RawData/FastQC_Reports/*") // Raw FastQC reports 
-    path("VV_INPUT/00-RawData/FastQC_Reports/*") // Unzipped Raw MultiQC report
-    path("VV_INPUT/00-RawData/FastQC_Reports/*") // Zipped Raw MultiQC report
-    path(dp_tools__NF_RCP)
+    path(runsheet)              // Runsheet
+    path("INPUT/raw_fastq/*")   // Raw reads
+    path("INPUT/raw_fastqc/*")  // Raw FastQC reports 
+    path("INPUT/raw_multiqc/*") // Unzipped Raw MultiQC report
 
   output:
-    path("Metadata/*_runsheet.csv"),                                                 emit: VVed_runsheet
     path("00-RawData/Fastq"),                                                        emit: VVed_raw_reads
     path("00-RawData/FastQC_Reports/*{_fastqc.html,_fastqc.zip}"),                   emit: VVed_raw_fastqc
-    path("00-RawData/FastQC_Reports/raw_multiqc${params.assay_suffix}_report"),     emit: VVed_raw_unzipped_multiqc_report
-    path("00-RawData/FastQC_Reports/raw_multiqc${params.assay_suffix}_report.zip"), emit: VVed_raw_zipped_multiqc_report
+    path("00-RawData/FastQC_Reports/raw_multiqc${params.assay_suffix}_report"),      emit: VVed_raw_unzipped_multiqc_report
+    //path("00-RawData/FastQC_Reports/raw_multiqc${params.assay_suffix}_report.zip"), emit: VVed_raw_zipped_multiqc_report
     path("VV_log.tsv"),                                                              optional: params.skipVV, emit: log
 
   script:
     """
-    # move from VV_INPUT to task directory
-    # This allows detection as output files for publishing
-    mv VV_INPUT/* . || true
-
-    # Run V&V unless user requests to skip V&V
-    if ${ !params.skipVV } ; then
-      dpt validation run ${dp_tools__NF_RCP} . Metadata/*_runsheet.csv \\
-                          --data-asset-key-sets  \\
-                            ${ meta.paired_end ? "'demuliplexed paired end raw data,qc reports for paired end raw data'" : "'demuliplexed single end raw data,qc reports for single end raw data'"} \\
-                          --run-components \\
-                            'Metadata,Raw Reads,Raw Reads By Sample' \\
-                          --max-flag-code ${ params.max_flag_code } \\
-                          --output VV_log.tsv
-    fi
+    mv INPUT/* . || true
+    vv.py --assay-type rnaseq \
+    --assay-suffix ${params.assay_suffix} \
+    --runsheet-path ${runsheet} \
+    --outdir ${publishdir} \
+    --paired-end ${meta.paired_end} \
+    --mode microbes \
+    --raw-fastq raw_fastq/ \
+    --raw-fastqc raw_fastqc/ \
+    --raw-multiqc raw_multiqc/ \
+    --run-components raw_reads
     """
 }