MD finally supports cram output (#2019)

* MD finally supports cram output

* publish crai file

* fix test output

* fix mv

* figure out if contains breaks stuff

* remove variable checksum

* add contains check

* remove variable checksum

Co-authored-by: Maxime U. Garcia <max.u.garcia@gmail.com>

modules/gatk4/markduplicates/main.nf

@@ -9,10 +9,14 @@ process GATK4_MARKDUPLICATES {
 
     input:
     tuple val(meta), path(bam)
+    path  fasta
+    path  fasta_fai
 
     output:
-    tuple val(meta), path("*.bam")    , emit: bam
-    tuple val(meta), path("*.bai")    , optional:true, emit: bai
+    tuple val(meta), path("*cram"),     emit: cram,  optional: true
+    tuple val(meta), path("*bam"),      emit: bam,   optional: true
+    tuple val(meta), path("*.crai"),    emit: crai,  optional: true
+    tuple val(meta), path("*.bai"),     emit: bai,   optional: true
     tuple val(meta), path("*.metrics"), emit: metrics
     path "versions.yml",                emit: versions
 
@@ -21,8 +25,9 @@ process GATK4_MARKDUPLICATES {
 
     script:
     def args = task.ext.args ?: ''
-    def prefix = task.ext.prefix ?: "${meta.id}"
+    prefix = task.ext.prefix ?: "${meta.id}"
     def input_list = bam.collect{"--INPUT $it"}.join(' ')
+    def reference = fasta ? "--REFERENCE_SEQUENCE ${fasta}" : ""
 
     def avail_mem = 3
     if (!task.memory) {
@@ -33,11 +38,17 @@ process GATK4_MARKDUPLICATES {
     """
     gatk --java-options "-Xmx${avail_mem}g" MarkDuplicates \\
         $input_list \\
-        --OUTPUT ${prefix}.bam \\
+        --OUTPUT ${prefix} \\
         --METRICS_FILE ${prefix}.metrics \\
         --TMP_DIR . \\
+        ${reference} \\
         $args
 
+
+    if  [[ ${prefix} == *.cram ]]; then
+        mv ${prefix}.bai ${prefix}.crai
+    fi
+
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
         gatk4: \$(echo \$(gatk --version 2>&1) | sed 's/^.*(GATK) v//; s/ .*\$//')

modules/gatk4/markduplicates/meta.yml

@@ -26,6 +26,14 @@ input:
       type: file
       description: Sorted BAM file
       pattern: "*.{bam}"
+  - fasta:
+      type: file
+      description: Fasta file
+      pattern: "*.{fasta}"
+  - fasta_fai:
+      type: file
+      description: Fasta index file
+      pattern: "*.{fai}"
 
 output:
   - meta:
@@ -41,6 +49,18 @@ output:
       type: file
       description: Marked duplicates BAM file
       pattern: "*.{bam}"
+  - cram:
+      type: file
+      description: Marked duplicates CRAM file
+      pattern: "*.{cram}"
+  - bai:
+      type: file
+      description: BAM index file
+      pattern: "*.{bam.bai}"
+  - crai:
+      type: file
+      description: CRAM index file
+      pattern: "*.{cram.crai}"
   - metrics:
       type: file
       description: Duplicate metrics file generated by GATK

tests/modules/gatk4/markduplicates/main.nf

@@ -3,13 +3,14 @@
 nextflow.enable.dsl = 2
 
 include { GATK4_MARKDUPLICATES } from '../../../../modules/gatk4/markduplicates/main.nf'
+include { GATK4_MARKDUPLICATES as GATK4_MARKDUPLICATES_CRAM } from '../../../../modules/gatk4/markduplicates/main.nf'
 
 workflow test_gatk4_markduplicates {
     input = [ [ id:'test', single_end:false ], // meta map
               file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true)
             ]
 
-    GATK4_MARKDUPLICATES ( input )
+    GATK4_MARKDUPLICATES ( input, [], [] )
 }
 
 workflow test_gatk4_markduplicates_multiple_bams {
@@ -18,5 +19,16 @@ workflow test_gatk4_markduplicates_multiple_bams {
                 file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_sorted_bam'], checkIfExists: true)
             ] ]
 
-    GATK4_MARKDUPLICATES ( input )
+    GATK4_MARKDUPLICATES ( input, [], [] )
+}
+
+workflow test_gatk4_markduplicates_multiple_cram_output {
+    input = [ [ id:'test', single_end:false ], // meta map
+              [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true),
+                file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_sorted_bam'], checkIfExists: true)
+            ] ]
+    fasta = file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
+    fai = file(params.test_data['homo_sapiens']['genome']['genome_fasta_fai'], checkIfExists: true)
+
+    GATK4_MARKDUPLICATES_CRAM ( input, fasta, fai )
 }

tests/modules/gatk4/markduplicates/nextflow.config

@@ -4,6 +4,12 @@ process {
 
     withName: GATK4_MARKDUPLICATES {
         ext.args = '--CREATE_INDEX true'
+        ext.prefix = { "${meta.id}.bam" }
+    }
+
+    withName: GATK4_MARKDUPLICATES_CRAM {
+        ext.args = '--CREATE_INDEX true'
+        ext.prefix = { "${meta.id}.cram" }
     }
 
 }

tests/modules/gatk4/markduplicates/test.yml

@@ -1,25 +1,40 @@
 - name: gatk4 markduplicates test_gatk4_markduplicates
-  command: nextflow run tests/modules/gatk4/markduplicates -entry test_gatk4_markduplicates -c tests/config/nextflow.config -c ./tests/modules/gatk4/markduplicates/nextflow.config
+  command: nextflow run ./tests/modules/gatk4/markduplicates -entry test_gatk4_markduplicates -c ./tests/config/nextflow.config  -c ./tests/modules/gatk4/markduplicates/nextflow.config
   tags:
     - gatk4
     - gatk4/markduplicates
   files:
     - path: output/gatk4/test.bai
-      md5sum: c8f7a9e426c768577f88f59cb1336bf3
+      md5sum: 4ac375b5288501c832bc3e4773f06d39
     - path: output/gatk4/test.bam
-      md5sum: 58533ddab47f7ac07f7b10e7f4aac234
-    - path: output/gatk4/test.metrics
-    - path: output/gatk4/versions.yml
+      md5sum: c1a0d25c5f65514e72aa490326557a88
+    - path: output/gatk4/test.bam.metrics
 
 - name: gatk4 markduplicates test_gatk4_markduplicates_multiple_bams
-  command: nextflow run tests/modules/gatk4/markduplicates -entry test_gatk4_markduplicates_multiple_bams -c tests/config/nextflow.config -c ./tests/modules/gatk4/markduplicates/nextflow.config
+  command: nextflow run ./tests/modules/gatk4/markduplicates -entry test_gatk4_markduplicates_multiple_bams -c ./tests/config/nextflow.config  -c ./tests/modules/gatk4/markduplicates/nextflow.config
   tags:
     - gatk4
     - gatk4/markduplicates
   files:
     - path: output/gatk4/test.bai
-      md5sum: bad71df9c876e72a5bc0a3e0fd755f92
+      md5sum: 832fbc6b3c3266d024fc7131ffd1b45d
     - path: output/gatk4/test.bam
-      md5sum: 112580c24b43331950f24f9adea30788
-    - path: output/gatk4/test.metrics
-    - path: output/gatk4/versions.yml
+      md5sum: 36909e80bde1aa547fae6382fd35764d
+    - path: output/gatk4/test.bam.metrics
+      contains:
+        - "testN"
+        - "2820"
+
+- name: gatk4 markduplicates test_gatk4_markduplicates_multiple_cram_output
+  command: nextflow run ./tests/modules/gatk4/markduplicates -entry test_gatk4_markduplicates_multiple_cram_output -c ./tests/config/nextflow.config  -c ./tests/modules/gatk4/markduplicates/nextflow.config
+  tags:
+    - gatk4
+    - gatk4/markduplicates
+  files:
+    - path: output/gatk4/test.cram
+    - path: output/gatk4/test.cram.crai
+      md5sum: 53a329f0828210e317ad81b0e6ca4bc5
+    - path: output/gatk4/test.cram.metrics
+      contains:
+        - "testN"
+        - "2820"