Merge pull request #172 from nf-core/ciriquant

Implement ciriquant quantification

conf/modules.config

@@ -389,7 +389,8 @@ process {
         ]
     }
 
-    withName: '.*:CIRIQUANT:MAIN' {
+    withName: '.*:BSJ_DETECTION:CIRIQUANT:MAIN' {
+        ext.args = "--no-gene"
         publishDir = [
             path: { "${params.outdir}/bsj_detection/tools/ciriquant/intermediates/${meta.id}" },
             mode: params.publish_dir_mode,
@@ -664,6 +665,7 @@ process {
 
     withName: 'COMBINE_COUNTS_PER_TOOL' {
         ext.args = "-f 1 -t -O"
+        maxRetries = 3
         publishDir = [
             path: { "${params.outdir}/bsj_detection/tools/${meta.id}" },
             mode: params.publish_dir_mode,
@@ -787,6 +789,7 @@ process {
 
     withName: '.*:ANNOTATE_(COMBINED|PER_SAMPLE|PER_SAMPLE_TOOL):ANNOTATE' {
         ext.prefix = { "${meta.id}.annotated" }
+        maxRetries = 3
     }
 
     withName: '.*:ANNOTATE_COMBINED:ANNOTATE' {
@@ -868,6 +871,7 @@ process {
     }
 
     withName: TRANSCRIPTOME {
+        ext.args = "-w"
         publishDir = [
             path: { "${params.outdir}/quantification/transcriptome" },
             mode: params.publish_dir_mode,
@@ -888,6 +892,62 @@ process {
         ]
     }
 
+    withName: '.*:QUANTIFICATION:CIRIQUANT:MAIN' {
+        publishDir = [
+            [
+                path: { "${params.outdir}/quantification/ciriquant/results/transcripts" },
+                mode: params.publish_dir_mode,
+                pattern: "*/gene/*_out.gtf",
+                saveAs: { filename -> filename.equals('versions.yml') ? null : filename.split('/').last() }
+            ],
+            [
+                path: { "${params.outdir}/quantification/ciriquant/results/genes" },
+                mode: params.publish_dir_mode,
+                pattern: "*/gene/*_genes.list",
+                saveAs: { filename -> filename.equals('versions.yml') ? null : filename.split('/').last() }
+            ],
+            [
+                path: { "${params.outdir}/quantification/ciriquant/results/circs" },
+                mode: params.publish_dir_mode,
+                pattern: "*/*.gtf",
+                saveAs: { filename -> filename.equals('versions.yml') ? null : filename.split('/').last() }
+            ]
+        ]
+    }
+
+    withName: '.*:QUANTIFICATION:CIRIQUANT:EXTRACT_CIRC' {
+        // PYGTFTK process that transforms a GTF file into circ_id, gene_id, count
+        ext.args = "-k circ_id,gene_id,score -s \\\t"
+        ext.suffix = "circ.tsv"
+        publishDir = [
+            path: { "${params.outdir}/quantification/ciriquant/circ" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+        ]
+    }
+
+    withName: '.*:QUANTIFICATION:CIRIQUANT:EXTRACT_GENES' {
+        ext.args = "-v FS='\\t' -v OFS='\\t' '{ print \$1, \$2, \$9 }'"
+        ext.suffix = "gene.tsv"
+        publishDir = [
+            path: { "${params.outdir}/quantification/ciriquant/gene" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+        ]
+    }
+
+    withName: ".*:CIRIQUANT:JOIN_(GENE|CIRC)" {
+        publishDir = [
+            path: { "${params.outdir}/quantification/ciriquant/combined" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+        ]
+    }
+
+    withName: ".*:CIRIQUANT:JOIN_CIRC" {
+        ext.metacols = "circ_id,gene_id"
+    }
+
     withName: PSIRC_INDEX {
         publishDir = [
             path: { "${params.outdir}/references/index/psirc" },
@@ -896,7 +956,7 @@ process {
         ]
     }
 
-    withName: PSIRC_QUANT {
+    withName: RUN_PSIRC_QUANT {
         publishDir = [
             path: { "${params.outdir}/quantification/samples/${meta.id}/psirc" },
             mode: params.publish_dir_mode,
@@ -928,7 +988,7 @@ process {
         ]
     }
 
-    withName: ".*:JOIN_(GENE|TX)_(COUNTS|TPM)" {
+    withName: ".*:PSIRC_QUANT:JOIN_(GENE|TX)_(COUNTS|TPM)" {
         ext.args = "-f 1,2 -t"
         label = "process_medium"
         maxRetries = 3
@@ -1049,4 +1109,28 @@ process {
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
     }
+
+    withName: CIRIQUANT_PREPDE {
+        publishDir = [
+            path: { "${params.outdir}/statistical_tests/ciriquant_de/ciriquant_prep" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+        ]
+    }
+
+    withName: STRINGTIE_PREPDE {
+        publishDir = [
+            path: { "${params.outdir}/statistical_tests/ciriquant_de/stringtie_prep" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+        ]
+    }
+
+    withName: CIRIQUANT_DE {
+        publishDir = [
+            path: { "${params.outdir}/statistical_tests/ciriquant_de" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+        ]
+    }
 }

modules.json

@@ -111,6 +111,11 @@
                         "git_sha": "cf3ed075695639b0a0924eb0901146df1996dc08",
                         "installed_by": ["modules"]
                     },
+                    "gffread": {
+                        "branch": "master",
+                        "git_sha": "6c996d7fbe0816dcbb68ce587ad5f873313682a1",
+                        "installed_by": ["modules"]
+                    },
                     "gnu/sort": {
                         "branch": "master",
                         "git_sha": "a3cc42943548378b726610f45bb5a79ab3f0b633",

modules/local/annotation/main.nf

@@ -1,6 +1,6 @@
 process ANNOTATION {
     tag "$meta.id"
-    label 'process_single'
+    label 'process_medium'
 
     conda "${moduleDir}/environment.yml"
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?

modules/local/circtest/circtest/main.nf

@@ -7,9 +7,8 @@ process CIRCTEST_CIRCTEST {
         'biocontainers/mulled-v2-c79b00aa4647c739dbe7e8480789d3ba67988f2e:0' }"
 
     input:
-    tuple val(meta) , path(circ_counts)
-    tuple val(meta2), path(gene_counts)
-    tuple val(meta3), path(phenotype)
+    tuple val(meta) , path(gene_counts), path(circ_counts)
+    tuple val(meta2), path(phenotype)
 
     output:
     tuple val(meta), path("${prefix}_summary.txt"), emit: summary

modules/local/circtest/prepare/main.nf

@@ -5,12 +5,10 @@ process CIRCTEST_PREPARE {
     container "biocontainers/r-base:4.2.1"
 
     input:
-    tuple val(meta), path(circ_counts)
-    tuple val(meta2), path(gene_counts)
+    tuple val(meta), path(gene_counts), path(circ_counts)
 
     output:
-    tuple val(meta), path('*_circs.tsv'), emit: circ_counts, optional: true
-    tuple val(meta), path('*_genes.tsv'), emit: gene_counts, optional: true
+    tuple val(meta), path('*_genes.tsv'), path('*_circs.tsv'), emit: counts, optional: true
 
     path "versions.yml"                 , emit: versions
 

modules/local/ciriquant/Dockerfile

@@ -0,0 +1,4 @@
+FROM community.wave.seqera.io/library/bioconductor-edger_bwa_hisat2_pysam_pruned:f14fb8726c7f0cf8
+
+# Install custom fork
+RUN pip install git+https://github.com/nictru/CIRIquant.git@e4916ca7b3370cef54d76ca162be64792d8c1b16

modules/local/ciriquant/main.nf → modules/local/ciriquant/ciriquant/main.nf

@@ -2,22 +2,21 @@ process CIRIQUANT {
     tag "$meta.id"
     label 'process_high'
 
-    conda "bioconda::ciriquant=1.1.2"
-    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/ciriquant:1.1.2--pyhdfd78af_2' :
-        'biocontainers/ciriquant:1.1.2--pyhdfd78af_2' }"
+    container "docker.io/nicotru/ciriquant:1.0.4"
 
     input:
     tuple val(meta), path(reads)
-    tuple val(meta2), path(gtf)
-    tuple val(meta3), path(fasta)
-    tuple val(meta4), path(bwa)
-    tuple val(meta5), path(hisat2)
+    tuple val(meta2), path(bed)
+    tuple val(meta3), path(gtf)
+    tuple val(meta4), path(fasta)
+    tuple val(meta5), path(bwa)
+    tuple val(meta6), path(hisat2)
 
     output:
-    tuple val(meta), path("${prefix}/${prefix}.gtf"), emit: gtf
-    path("${prefix}")                               , emit: results
-    path "versions.yml"                             , emit: versions
+    tuple val(meta), path("${prefix}/${prefix}.gtf")            , emit: gtf
+    tuple val(meta), path("${prefix}/gene/${prefix}_genes.list"), emit: gene_list, optional: true
+    tuple val(meta), path("${prefix}/gene/${prefix}_out.gtf")   , emit: gene_gtf, optional: true
+    path "versions.yml"                                         , emit: versions
 
     when:
     task.ext.when == null || task.ext.when
@@ -28,6 +27,8 @@ process CIRIQUANT {
     def VERSION = '2.1.0'
     def strandedness = meta.strandedness ?: 'auto'
     def library_type = strandedness == 'auto' ? '' : strandedness == 'unstranded' ? '-l 0' : strandedness == 'forward' ? '-l 1' : '-l 2'
+    def reads_string = meta.single_end ? "-r ${reads}" : "-1 ${reads[0]} -2 ${reads[1]}"
+    def bed_string = bed ? "--bed ${bed}" : ''
     """
     BWA=`which bwa`
     HISAT2=`which hisat2`
@@ -44,10 +45,9 @@ process CIRIQUANT {
 
     CIRIquant \\
         -t ${task.cpus} \\
-        -1 ${reads[0]} \\
-        -2 ${reads[1]} \\
+        ${reads_string} \\
+        ${bed_string} \\
         --config config.yml \\
-        --no-gene \\
         -o ${prefix} \\
         -p ${prefix} \\
         ${library_type} \\

modules/local/ciriquant/de/main.nf

@@ -0,0 +1,35 @@
+process CIRIQUANT_DE {
+    tag "$meta.id"
+    label 'process_high'
+
+    container "docker.io/nicotru/ciriquant:1.0.4"
+
+    input:
+    tuple val(meta), path(library), path(expression), path(gene)
+
+    output:
+    tuple val(meta), path("${circ_path}"), emit: circ
+    tuple val(meta), path("${gene_path}"), emit: gene
+    path "versions.yml", emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    prefix = task.ext.prefix ?: "${meta.id}"
+    circ_path = "${prefix}.circ.csv"
+    gene_path = "${prefix}.gene.csv"
+    """
+    CIRI_DE_replicate \\
+        --lib ${library} \\
+        --bsj ${expression} \\
+        --gene ${gene} \\
+        --out ${circ_path} \\
+        --out2 ${gene_path}
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        ciriquant: \$(echo \$(CIRIquant --version 2>&1) | sed 's/CIRIquant //g' )
+    END_VERSIONS
+    """
+}

modules/local/ciriquant/prepde/main.nf

@@ -0,0 +1,42 @@
+process CIRIQUANT_PREPDE {
+    tag "$meta.id"
+    label 'process_high'
+
+    container "docker.io/nicotru/ciriquant:1.0.4"
+
+    input:
+    tuple val(meta), val(samples), path(gtfs), val(conditions)
+
+    output:
+    tuple val(meta), path("${prefix}_library.tsv")   , emit: library
+    tuple val(meta), path("${prefix}_annotation.tsv"), emit: annotation
+    tuple val(meta), path("${prefix}_expression.tsv"), emit: expression
+    tuple val(meta), path("${prefix}_ratio.tsv")     , emit: gene
+    path "versions.yml"                              , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    prefix = task.ext.prefix ?: "${meta.id}"
+    samplesheet = [samples, gtfs, conditions]
+        .transpose()
+        .collect{ sample, gtf, condition ->
+            "${sample}\\t${gtf}\\t${condition}" }.join('\\n')
+    """
+    echo -e "${samplesheet}" > samples.txt
+
+    prep_CIRIquant -i samples.txt \\
+        --lib ${prefix}_library.tsv \\
+        --circ ${prefix}_annotation.tsv \\
+        --bsj ${prefix}_expression.tsv \\
+        --ratio ${prefix}_ratio.tsv \\
+        ${args}
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        ciriquant: \$(echo \$(CIRIquant --version 2>&1) | sed 's/CIRIquant //g' )
+    END_VERSIONS
+    """
+}

modules/local/majority_vote/environment.yml

@@ -0,0 +1,7 @@
+name: annotation
+channels:
+  - conda-forge
+  - bioconda
+dependencies:
+  - conda-forge::polars=1.5.0
+  - conda-forge::pyyaml=6.0.2

modules/local/majority_vote/main.nf

@@ -2,15 +2,14 @@ process MAJORITY_VOTE {
     tag "$meta.id"
     label 'process_medium'
 
-    conda "bioconda::pandas=1.5.2"
+    conda "${moduleDir}/environment.yml"
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-            'https://depot.galaxyproject.org/singularity/pandas:1.5.2' :
-            'biocontainers/pandas:1.5.2' }"
+            'oras://community.wave.seqera.io/library/polars_pyyaml:962a0cf7480258c7' :
+            'community.wave.seqera.io/library/polars_pyyaml:ad93db0d7bcd508e' }"
 
     input:
     tuple val(meta), path(bindingsites)
 
-
     output:
     tuple val(meta), path("${meta.id}.majority.tsv"), emit: tsv
     tuple val(meta), path("${meta.id}.targets.tsv") , emit: targets
@@ -20,6 +19,7 @@ process MAJORITY_VOTE {
     task.ext.when == null || task.ext.when
 
     script:
+    min_tools = params.mirna_tool_filter
     template 'majority.py'
 
     stub: