controlfreec significance (#1451)

* controlfreec significance

* move freec files to own subfolder

* Fix meta.yml naming

* Fix meta.yml naming

* Fix linting

* Forgot to refactor

* forgot more refactoring

* Too much refactoring on output paths

* Too little refactoring here

* update checksum

modules/controlfreec/assesssignificance/main.nf

@@ -0,0 +1,30 @@
+process CONTROLFREEC_ASSESSSIGNIFICANCE {
+    tag "$meta.id"
+    label 'process_low'
+
+    conda (params.enable_conda ? "bioconda::control-freec=11.6" : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/control-freec:11.6--h1b792b2_1':
+        'quay.io/biocontainers/control-freec:11.6--h1b792b2_1' }"
+
+    input:
+    tuple val(meta), path(cnvs), path(ratio)
+
+    output:
+    tuple val(meta), path("*.p.value.txt"), emit: bam
+    path "versions.yml"                   , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    """
+    cat /usr/local/bin/assess_significance.R | R --slave --args ${cnvs} ${ratio}
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        controlfreec: \$(echo \$(freec -version 2>&1) | sed 's/^.*Control-FREEC  //; s/:.*\$//' | sed -e "s/Control-FREEC v//g" )
+    END_VERSIONS
+    """
+}

modules/controlfreec/assesssignificance/meta.yml

@@ -0,0 +1,50 @@
+name: controlfreec_assesssignificance
+description: Add both Wilcoxon test and Kolmogorov-Smirnov test p-values to each CNV output of FREEC
+keywords:
+  - cna
+  - cnv
+  - somatic
+  - single
+  - tumor-only
+tools:
+  - controlfreec/assesssignificance:
+      description: Copy number and genotype annotation from whole genome and whole exome sequencing data.
+      homepage: http://boevalab.inf.ethz.ch/FREEC
+      documentation: http://boevalab.inf.ethz.ch/FREEC/tutorial.html
+      tool_dev_url: https://github.com/BoevaLab/FREEC/
+      doi: "10.1093/bioinformatics/btq635"
+      licence: ["GPL >=2"]
+
+input:
+  # Only when we have meta
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - cnvs:
+      type: file
+      description: _CNVs file generated by FREEC
+      pattern: "*._CNVs"
+  - ratio:
+      type: file
+      description: ratio file generated by FREEC
+      pattern: "*.ratio.txt"
+
+output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+  - p_value_txt:
+      type: file
+      description: CNV file containing p_values for each call
+      pattern: "*.p.value.txt"
+
+authors:
+  - "@FriederikeHanssen"

modules/controlfreec/freec/main.nf

@@ -1,4 +1,4 @@
-process CONTROLFREEC {
+process CONTROLFREEC_FREEC {
     tag "$meta.id"
     label 'process_low'
 

modules/controlfreec/freec/meta.yml

@@ -1,4 +1,4 @@
-name: controlfreec
+name: controlfreec_freec
 description: Copy number and genotype annotation from whole genome and whole exome sequencing data
 keywords:
   - cna
@@ -7,7 +7,7 @@ keywords:
   - single
   - tumor-only
 tools:
-  - controlfreec:
+  - controlfreec/freec:
       description: Copy number and genotype annotation from whole genome and whole exome sequencing data.
       homepage: http://boevalab.inf.ethz.ch/FREEC
       documentation: http://boevalab.inf.ethz.ch/FREEC/tutorial.html

tests/config/pytest_modules.yml

@@ -419,9 +419,13 @@ cnvkit/batch:
   - modules/cnvkit/batch/**
   - tests/modules/cnvkit/batch/**
 
-controlfreec:
+controlfreec/assesssignificance:
-  - modules/controlfreec/**
+  - modules/controlfreec/assesssignificance/**
-  - tests/modules/controlfreec/**
+  - tests/modules/controlfreec/assesssignificance/**
+
+controlfreec/freec:
+  - modules/controlfreec/freec/**
+  - tests/modules/controlfreec/freec/**
 
 cooler/cload:
   - modules/cooler/cload/**

tests/modules/controlfreec/assesssignificance/main.nf

@@ -0,0 +1,42 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { CONTROLFREEC_ASSESSSIGNIFICANCE } from '../../../../modules/controlfreec/assesssignificance/main.nf'
+include { CONTROLFREEC_FREEC              } from '../../../../modules/controlfreec/freec/main.nf'
+include { UNTAR                           }        from '../../../../modules/untar/main.nf'
+
+workflow test_controlfreec_assesssignificance {
+
+    input = [
+        [ id:'test', single_end:false, sex:'XX' ], // meta map
+        file(params.test_data['homo_sapiens']['illumina']['test_mpileup'], checkIfExists: true),
+        file(params.test_data['homo_sapiens']['illumina']['test2_mpileup'], checkIfExists: true),
+        [],[],[],[]
+    ]
+
+    fasta = file(params.test_data['homo_sapiens']['genome']['genome_21_fasta'], checkIfExists: true)
+    fai = file(params.test_data['homo_sapiens']['genome']['genome_21_fasta_fai'], checkIfExists: true)
+
+    dbsnp = file(params.test_data['homo_sapiens']['genome']['dbsnp_138_hg38_21_vcf_gz'], checkIfExists: true)
+    dbsnp_tbi = file(params.test_data['homo_sapiens']['genome']['dbsnp_138_hg38_21_vcf_gz_tbi'], checkIfExists: true)
+
+    chrfiles = [ [], file(params.test_data['homo_sapiens']['genome']['genome_21_chromosomes_dir'], checkIfExists: true) ]
+    target_bed = file(params.test_data['homo_sapiens']['genome']['genome_21_multi_interval_bed'], checkIfExists: true)
+
+    UNTAR(chrfiles)
+    CONTROLFREEC_FREEC (input,
+                        fasta,
+                        fai,
+                        [],
+                        dbsnp,
+                        dbsnp_tbi,
+                        UNTAR.out.untar.map{ it[1] },
+                        [],
+                        target_bed,
+                        []
+                        )
+
+    sig_in = CONTROLFREEC_FREEC.out.CNV.join(CONTROLFREEC_FREEC.out.ratio)
+    CONTROLFREEC_ASSESSSIGNIFICANCE ( sig_in )
+}

tests/modules/controlfreec/assesssignificance/nextflow.config

@@ -2,7 +2,7 @@ process {
 
     publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
 
-    withName:CONTROLFREEC{
+    withName:CONTROLFREEC_FREEC{
         ext.args = { [
                     "sample":[
                         inputformat: 'pileup',

tests/modules/controlfreec/assesssignificance/test.yml

@@ -0,0 +1,10 @@
+- name: controlfreec assesssignificance test_controlfreec_assesssignificance
+  command: nextflow run tests/modules/controlfreec/assesssignificance -entry test_controlfreec_assesssignificance -c tests/config/nextflow.config
+  tags:
+    - controlfreec/assesssignificance
+    - controlfreec
+  files:
+    - path: output/controlfreec/test2.mpileup.gz_CNVs.p.value.txt
+      md5sum: 44e23b916535fbc1a3f47b57fad292df
+    - path: output/controlfreec/versions.yml
+      md5sum: 0aa42fed10d61e4570fe1e0e83ffe932

tests/modules/controlfreec/freec/main.nf

@@ -2,9 +2,10 @@
 
 nextflow.enable.dsl = 2
 
-include { CONTROLFREEC } from '../../../modules/controlfreec/main.nf'
+include { CONTROLFREEC_FREEC } from '../../../../modules/controlfreec/freec/main.nf'
-include { UNTAR }        from '../../../modules/untar/main.nf'
+include { UNTAR              }        from '../../../../modules/untar/main.nf'
-workflow test_controlfreec {
+
+workflow test_controlfreec_freec {
 
     input = [
         [ id:'test', single_end:false, sex:'XX' ], // meta map
@@ -23,15 +24,15 @@ workflow test_controlfreec {
     target_bed = file(params.test_data['homo_sapiens']['genome']['genome_21_multi_interval_bed'], checkIfExists: true)
 
     UNTAR(chrfiles)
-    CONTROLFREEC (  input,
+    CONTROLFREEC_FREEC (input,
-                            fasta,
+                        fasta,
-                            fai,
+                        fai,
-                            [],
+                        [],
-                            dbsnp,
+                        dbsnp,
-                            dbsnp_tbi,
+                        dbsnp_tbi,
-                            UNTAR.out.untar.map{ it[1] },
+                        UNTAR.out.untar.map{ it[1] },
-                            [],
+                        [],
-                            target_bed,
+                        target_bed,
-                            []
+                        []
                         )
 }

tests/modules/controlfreec/freec/nextflow.config

@@ -0,0 +1,26 @@
+process {
+
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+
+    withName:CONTROLFREEC_FREEC{
+        ext.args = { [
+                    "sample":[
+                        inputformat: 'pileup',
+                        mateorientation: 'FR'
+                    ],
+                    "general" :[
+                        bedgraphoutput: "TRUE",
+                        noisydata: "TRUE",
+                        minexpectedgc: "0",
+                        readcountthreshold: "1",
+                        sex: meta.sex,
+                        window: "10",
+                        ],
+                    "control":[
+                        inputformat: "pileup",
+                        mateorientation: "FR"
+                    ]
+                    ]
+        }
+    }
+}

tests/modules/controlfreec/freec/test.yml

@@ -1,7 +1,8 @@
-- name: controlfreec test_controlfreec
+- name: controlfreec test_controlfreec_freec
-  command: nextflow run tests/modules/controlfreec -entry test_controlfreec -c tests/config/nextflow.config
+  command: nextflow run tests/modules/controlfreec/freec -entry test_controlfreec_freec -c tests/config/nextflow.config
   tags:
     - controlfreec
+    - controlfreec/freec
   files:
     - path: output/controlfreec/config.txt
     - path: output/controlfreec/test.mpileup.gz_control.cpn
@@ -19,4 +20,4 @@
     - path: output/controlfreec/test2.mpileup.gz_sample.cpn
       md5sum: c80dad58a77b1d7ba6d273999f4b4b4b
     - path: output/controlfreec/versions.yml
-      md5sum: ff93f6466d4686aab708425782c6c848
+      md5sum: 3ab250a2ab3be22628124c7c65324651