CNVkit: Make targets file optional when running in WGS mode (#1030)

* Make targets.bed optional when running in wgs mode

* added test for cram

* Update test_data_config with new reference.cnn

* Update main.nf to allow tumor-only running

Still need a unit-test for this. Almost ready, but needs this file as input https://github.com/nf-core/test-datasets/blob/modules/data/generic/cnn/reference.cnn

* re-writing previous changes, but now it wont crash the entire CI-setup

* fixing overlooked merge conflict

* last overlooked merge-conflict

* move all files to batch subfolder

* adding an optional input for a reference file (needed when running germline and tumoronly)

* minor typo

* update meta.yml

* aligning code, renaming cnvkit to cnvkit_batch, renaming tumorbam to tumor, normalbam to normal

* Update pytest_modules.yml

Co-authored-by: EC2 Default User <ec2-user@ip-172-31-21-198.us-west-2.compute.internal>
Co-authored-by: Lasse Folkersen <lassefolkersen@gmail.com>
Co-authored-by: Robert A. Petit III <robbie.petit@gmail.com>
Co-authored-by: Harshil Patel <drpatelh@users.noreply.github.com>

modules/cnvkit/batch/main.nf

@@ -4,7 +4,7 @@ include { initOptions; saveFiles; getSoftwareName; getProcessName } from './func
 params.options = [:]
 options        = initOptions(params.options)
 
-process CNVKIT {
+process CNVKIT_BATCH {
     tag "$meta.id"
     label 'process_low'
     publishDir "${params.outdir}",
@@ -19,25 +19,40 @@ process CNVKIT {
     }
 
     input:
-    tuple val(meta), path(tumourbam), path(normalbam)
+    tuple val(meta), path(tumor), path(normal)
     path  fasta
-    path  targetfile
+    path  targets
+    path  reference
 
     output:
     tuple val(meta), path("*.bed"), emit: bed
-    tuple val(meta), path("*.cnn"), emit: cnn
-    tuple val(meta), path("*.cnr"), emit: cnr
-    tuple val(meta), path("*.cns"), emit: cns
+    tuple val(meta), path("*.cnn"), emit: cnn, optional: true
+    tuple val(meta), path("*.cnr"), emit: cnr, optional: true
+    tuple val(meta), path("*.cns"), emit: cns, optional: true
     path "versions.yml"           , emit: versions
 
     script:
+    normal_args = normal ? "--normal $normal" : ""
+    fasta_args = fasta ? "--fasta $fasta" : ""
+    reference_args = reference ? "--reference $reference" : ""
+
+    def target_args = ""
+    if (options.args.contains("--method wgs") || options.args.contains("-m wgs")) {
+        target_args = targets ? "--targets $targets" : ""
+    }
+    else {
+        target_args = "--targets $targets"
+    }
+
     """
     cnvkit.py \\
         batch \\
-        $tumourbam \\
-        --normal $normalbam\\
-        --fasta $fasta \\
-        --targets $targetfile \\
+        $tumor \\
+        $normal_args \\
+        $fasta_args \\
+        $reference_args \\
+        $target_args \\
+        --processes ${task.cpus} \\
         $options.args
 
     cat <<-END_VERSIONS > versions.yml

modules/cnvkit/batch/meta.yml

@@ -1,4 +1,4 @@
-name: cnvkit
+name: cnvkit_batch
 description: Copy number variant detection from high-throughput sequencing data
 keywords:
   - bam
@@ -38,14 +38,14 @@ input:
       description: |
         Groovy Map containing sample information
         e.g. [ id:'test', single_end:false ]
-  - tumourbam:
+  - tumour:
       type: file
       description: |
-        Input tumour sample bam file
-  - normalbam:
+        Input tumour sample bam file (or cram)
+  - normal:
       type: file
       description: |
-        Input normal sample bam file
+        Input normal sample bam file (or cram)
   - fasta:
       type: file
       description: |
@@ -54,6 +54,10 @@ input:
       type: file
       description: |
         Input target bed file
+  - reference:
+      type: file
+      description: |
+        Input reference cnn-file (only for germline and tumor-only running)
 output:
   - meta:
       type: map
@@ -85,4 +89,5 @@ authors:
     - "@KevinMenden"
     - "@MaxUlysse"
     - "@drpatelh"
-
+    - "@fbdtemme"
+    - "@lassefolkersen"

tests/config/pytest_modules.yml

@@ -294,9 +294,9 @@ cmseq/polymut:
   - modules/cmseq/polymut/**
   - tests/modules/cmseq/polymut/**
 
-cnvkit:
-  - modules/cnvkit/**
-  - tests/modules/cnvkit/**
+cnvkit/batch:
+  - modules/cnvkit/batch/**
+  - tests/modules/cnvkit/batch/**
 
 cooler/digest:
   - modules/cooler/digest/**

tests/config/test_data.config

@@ -263,6 +263,9 @@ params {
             'txt' {
                 hello = "${test_data_dir}/generic/txt/hello.txt"
             }
+            'cnn' {
+                 reference = "${test_data_dir}/generic/cnn/reference.cnn"
+            }
             'cooler'{
                 test_pairix_pair_gz                           = "${test_data_dir}/genomics/homo_sapiens/cooler/cload/hg19/hg19.GM12878-MboI.pairs.subsample.blksrt.txt.gz"
                 test_pairix_pair_gz_px2                       = "${test_data_dir}/genomics/homo_sapiens/cooler/cload/hg19/hg19.GM12878-MboI.pairs.subsample.blksrt.txt.gz.px2"

tests/modules/cnvkit/batch/main.nf

@@ -0,0 +1,64 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { CNVKIT_BATCH as CNVKIT_HYBRID }      from '../../../../modules/cnvkit/batch/main.nf' addParams( options: [ 'args': '--output-reference reference.cnn' ] )
+include { CNVKIT_BATCH as CNVKIT_WGS }         from '../../../../modules/cnvkit/batch/main.nf' addParams( options: [ 'args': '--output-reference reference.cnn --method wgs' ] )
+include { CNVKIT_BATCH as CNVKIT_TUMORONLY }   from '../../../../modules/cnvkit/batch/main.nf' addParams( options: [ 'args': '--method wgs' ] )
+
+
+workflow test_cnvkit_hybrid {
+    tumor     = file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true)
+    normal    = file(params.test_data['sarscov2']['illumina']['test_single_end_sorted_bam'], checkIfExists: true)
+
+    input     = [ [ id:'test' ], // meta map
+                tumor,
+                normal
+              ]
+    fasta     = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+    targets   = file(params.test_data['sarscov2']['genome']['baits_bed'], checkIfExists: true)
+
+    CNVKIT_HYBRID ( input, fasta, targets, [] )
+}
+
+workflow test_cnvkit_wgs {
+    tumor     = file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_sorted_bam'], checkIfExists: true) 
+    normal    = file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true)
+
+    input     = [ [ id:'test'], // meta map
+                tumor,
+                normal
+                ]
+    fasta     = file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
+
+    CNVKIT_WGS ( input, fasta, [], [] )
+}
+
+
+workflow test_cnvkit_cram {
+    tumor     = file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_sorted_bam'], checkIfExists: true)
+    normal    = file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true)
+
+    input     = [ [ id:'test'], // meta map
+                tumor,
+                normal
+                ]
+    fasta     = file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
+
+    CNVKIT_WGS ( input, fasta, [], [] )
+}
+
+
+
+workflow test_cnvkit_tumoronly {
+    tumor     = file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_sorted_bam'], checkIfExists: true)
+
+    input     = [ [ id:'test'], // meta map
+                tumor,
+                [ ]
+            ]
+    fasta     = file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
+    reference = file(params.test_data['generic']['cnn']['reference'], checkIfExists: true)
+
+    CNVKIT_TUMORONLY ( input, [], [], reference )
+}

tests/modules/cnvkit/batch/test.yml

@@ -0,0 +1,101 @@
+- name: cnvkit batch test_cnvkit_hybrid
+  command: nextflow run tests/modules/cnvkit/batch -entry test_cnvkit_hybrid -c tests/config/nextflow.config
+  tags:
+    - cnvkit/batch
+    - cnvkit
+  files:
+    - path: output/cnvkit/baits.antitarget.bed
+      md5sum: d41d8cd98f00b204e9800998ecf8427e
+    - path: output/cnvkit/baits.target.bed
+      md5sum: 26d25ff2d6c45b6d92169b3559c6acdb
+    - path: output/cnvkit/reference.cnn
+      md5sum: ac99c1ad8b917b96ae15119146c91ab9
+    - path: output/cnvkit/test.paired_end.sorted.antitargetcoverage.cnn
+      md5sum: 203caf8cef6935bb50b4138097955cb8
+    - path: output/cnvkit/test.paired_end.sorted.bintest.cns
+      md5sum: 6544d979475def8a9f69ba42a985668d
+    - path: output/cnvkit/test.paired_end.sorted.call.cns
+      md5sum: f2ca59b4d50b0c317adc526c1b99b622
+    - path: output/cnvkit/test.paired_end.sorted.cnr
+      md5sum: 7e37d73ab604dbc3fe4ebb56aca9bdc3
+    - path: output/cnvkit/test.paired_end.sorted.cns
+      md5sum: 060af1aa637ed51812af19bcce24fcfe
+    - path: output/cnvkit/test.paired_end.sorted.targetcoverage.cnn
+      md5sum: 3fe80b6013ffc3e9968345e810158215
+    - path: output/cnvkit/test.single_end.sorted.antitargetcoverage.cnn
+      md5sum: 203caf8cef6935bb50b4138097955cb8
+    - path: output/cnvkit/test.single_end.sorted.targetcoverage.cnn
+      md5sum: aa8a018b1d4d1e688c9f9f6ae01bf4d7
+
+- name: cnvkit batch test_cnvkit_wgs
+  command: nextflow run tests/modules/cnvkit/batch -entry test_cnvkit_wgs -c tests/config/nextflow.config
+  tags:
+    - cnvkit/batch
+    - cnvkit
+  files:
+    - path: output/cnvkit/genome.antitarget.bed
+      md5sum: d41d8cd98f00b204e9800998ecf8427e
+    - path: output/cnvkit/genome.bed
+      md5sum: 87a15eb9c2ff20ccd5cd8735a28708f7
+    - path: output/cnvkit/genome.target.bed
+      md5sum: a13353ae9c8405e701390c069255bbd2
+    - path: output/cnvkit/reference.cnn
+      md5sum: 05c6211e0179885b8a83e44fd21d5f86
+    - path: output/cnvkit/test.paired_end.sorted.antitargetcoverage.cnn
+      md5sum: 203caf8cef6935bb50b4138097955cb8
+    - path: output/cnvkit/test.paired_end.sorted.targetcoverage.cnn
+      md5sum: ff526714696aa49bdc1dc8d00d965266
+    - path: output/cnvkit/test2.paired_end.sorted.antitargetcoverage.cnn
+      md5sum: 203caf8cef6935bb50b4138097955cb8
+    - path: output/cnvkit/test2.paired_end.sorted.bintest.cns
+      md5sum: 6544d979475def8a9f69ba42a985668d
+    - path: output/cnvkit/test2.paired_end.sorted.call.cns
+      md5sum: f6de754c34f780e6befee5b3ff0893f8
+    - path: output/cnvkit/test2.paired_end.sorted.cnr
+      md5sum: 80318d06c6b095945a0fb0e85e887cbc
+    - path: output/cnvkit/test2.paired_end.sorted.cns
+      md5sum: 76afa47afc4bd5de35aee8fdb54d3d3a
+    - path: output/cnvkit/test2.paired_end.sorted.targetcoverage.cnn
+      md5sum: 6ae6b3fce7299eedca6133d911c38fe1
+
+- name: cnvkit batch test_cnvkit_cram
+  command: nextflow run tests/modules/cnvkit/batch -entry test_cnvkit_cram -c tests/config/nextflow.config
+  tags:
+    - cnvkit/batch
+    - cnvkit
+  files:
+    - path: output/cnvkit/genome.antitarget.bed
+      md5sum: d41d8cd98f00b204e9800998ecf8427e
+    - path: output/cnvkit/genome.bed
+      md5sum: 87a15eb9c2ff20ccd5cd8735a28708f7
+    - path: output/cnvkit/genome.target.bed
+      md5sum: a13353ae9c8405e701390c069255bbd2
+    - path: output/cnvkit/reference.cnn
+      md5sum: 05c6211e0179885b8a83e44fd21d5f86
+    - path: output/cnvkit/test.paired_end.sorted.antitargetcoverage.cnn
+      md5sum: 203caf8cef6935bb50b4138097955cb8
+    - path: output/cnvkit/test.paired_end.sorted.targetcoverage.cnn
+      md5sum: ff526714696aa49bdc1dc8d00d965266
+    - path: output/cnvkit/test2.paired_end.sorted.antitargetcoverage.cnn
+      md5sum: 203caf8cef6935bb50b4138097955cb8
+    - path: output/cnvkit/test2.paired_end.sorted.bintest.cns
+      md5sum: 6544d979475def8a9f69ba42a985668d
+    - path: output/cnvkit/test2.paired_end.sorted.call.cns
+      md5sum: f6de754c34f780e6befee5b3ff0893f8
+    - path: output/cnvkit/test2.paired_end.sorted.cnr
+      md5sum: 80318d06c6b095945a0fb0e85e887cbc
+    - path: output/cnvkit/test2.paired_end.sorted.cns
+      md5sum: 76afa47afc4bd5de35aee8fdb54d3d3a
+    - path: output/cnvkit/test2.paired_end.sorted.targetcoverage.cnn
+      md5sum: 6ae6b3fce7299eedca6133d911c38fe1
+
+- name: cnvkit batch test_cnvkit_tumoronly
+  command: nextflow run tests/modules/cnvkit/batch -entry test_cnvkit_tumoronly -c tests/config/nextflow.config
+  tags:
+    - cnvkit/batch
+    - cnvkit
+  files:
+    - path: output/cnvkit/reference.antitarget-tmp.bed
+      md5sum: d41d8cd98f00b204e9800998ecf8427e
+    - path: output/cnvkit/reference.target-tmp.bed
+      md5sum: 26d25ff2d6c45b6d92169b3559c6acdb

tests/modules/cnvkit/main.nf

@@ -1,19 +0,0 @@
-#!/usr/bin/env nextflow
-
-nextflow.enable.dsl = 2
-
-include { CNVKIT } from '../../../modules/cnvkit/main.nf' addParams( options: [ 'args': '--output-reference reference.cnn' ] )
-
-workflow test_cnvkit {
-    tumourbam = file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true)
-    normalbam = file(params.test_data['sarscov2']['illumina']['test_single_end_sorted_bam'], checkIfExists: true)
-    
-    input = [ [ id:'test' ], // meta map
-              tumourbam, 
-              normalbam 
-            ]
-    fasta   = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
-    targets = file(params.test_data['sarscov2']['genome']['baits_bed'], checkIfExists: true)
-
-    CNVKIT ( input, fasta, targets )
-}

tests/modules/cnvkit/test.yml

@@ -1,27 +0,0 @@
-- name: cnvkit
-  command: nextflow run ./tests/modules/cnvkit/ -entry test_cnvkit -c tests/config/nextflow.config
-  tags:
-    - cnvkit
-  files:
-    - path: output/cnvkit/baits.target.bed
-      md5sum: 26d25ff2d6c45b6d92169b3559c6acdb
-    - path: output/cnvkit/baits.antitarget.bed
-      md5sum: d41d8cd98f00b204e9800998ecf8427e
-    - path: output/cnvkit/reference.cnn
-      md5sum: ac99c1ad8b917b96ae15119146c91ab9
-    - path: output/cnvkit/test.paired_end.sorted.targetcoverage.cnn
-      md5sum: 3fe80b6013ffc3e9968345e810158215
-    - path: output/cnvkit/test.paired_end.sorted.antitargetcoverage.cnn
-      md5sum: 203caf8cef6935bb50b4138097955cb8
-    - path: output/cnvkit/test.single_end.sorted.targetcoverage.cnn
-      md5sum: aa8a018b1d4d1e688c9f9f6ae01bf4d7
-    - path: output/cnvkit/test.single_end.sorted.antitargetcoverage.cnn
-      md5sum: 203caf8cef6935bb50b4138097955cb8
-    - path: output/cnvkit/test.paired_end.sorted.cnr
-      md5sum: 7e37d73ab604dbc3fe4ebb56aca9bdc3
-    - path: output/cnvkit/test.paired_end.sorted.cns
-      md5sum: 060af1aa637ed51812af19bcce24fcfe
-    - path: output/cnvkit/test.paired_end.sorted.bintest.cns
-      md5sum: 6544d979475def8a9f69ba42a985668d
-    - path: output/cnvkit/test.paired_end.sorted.call.cns
-      md5sum: f2ca59b4d50b0c317adc526c1b99b622