Manta consistancy (#1407)

* update tests

* update

* update

* make the manta inputs consistant for germline/somatic/tumoronly

* match chromosomes to cram file (chr21)

* undo genotypegvfs

* undo genotypegvfs

* update manta input structure to match strelka. tuple(relevant input), path(fasta), path(fai)

* update tests

* fix typos

* fix typos

Co-authored-by: Smith Nicholas <smith@in.tum.de>

modules/manta/germline/main.nf

@@ -8,10 +8,9 @@ process MANTA_GERMLINE {
         'quay.io/biocontainers/manta:1.6.0--h9ee0642_1' }"
 
     input:
-    tuple val(meta), path(input), path(index)
+    tuple val(meta), path(input), path(index), path(target_bed), path(target_bed_tbi)
     path fasta
     path fasta_fai
-    tuple path(target_bed), path(target_bed_tbi)
 
 
     output:

modules/manta/germline/meta.yml

@@ -31,14 +31,6 @@ input:
       type: file
       description: BAM/CRAM/SAM index file. For joint calling use a list of files.
       pattern: "*.{bai,crai,sai}"
-  - fasta:
-      type: file
-      description: Genome reference FASTA file
-      pattern: "*.{fa,fasta}"
-  - fasta_fai:
-      type: file
-      description: Genome reference FASTA index file
-      pattern: "*.{fa.fai,fasta.fai}"
   - target_bed:
       type: file
       description: BED file containing target regions for variant calling
@@ -47,6 +39,14 @@ input:
       type: file
       description: Index for BED file containing target regions for variant calling
       pattern: "*.{bed.tbi}"
+  - fasta:
+      type: file
+      description: Genome reference FASTA file
+      pattern: "*.{fa,fasta}"
+  - fasta_fai:
+      type: file
+      description: Genome reference FASTA index file
+      pattern: "*.{fa.fai,fasta.fai}"
 
 output:
   - meta:

modules/manta/somatic/main.nf

@@ -8,11 +8,9 @@ process MANTA_SOMATIC {
         'quay.io/biocontainers/manta:1.6.0--h9ee0642_1' }"
 
     input:
-    tuple val(meta), path(input_normal), path(input_index_normal), path(input_tumor), path(input_index_tumor)
+    tuple val(meta), path(input_normal), path(input_index_normal), path(input_tumor), path(input_index_tumor), path(target_bed), path(target_bed_tbi)
     path fasta
     path fai
-    path target_bed
-    path target_bed_tbi
 
     output:
     tuple val(meta), path("*.candidate_small_indels.vcf.gz")     , emit: candidate_small_indels_vcf

modules/manta/somatic/meta.yml

@@ -39,14 +39,6 @@ input:
       type: file
       description: BAM/CRAM/SAM index file
       pattern: "*.{bai,crai,sai}"
-  - fasta:
-      type: file
-      description: Genome reference FASTA file
-      pattern: "*.{fa,fasta}"
-  - fai:
-      type: file
-      description: Genome reference FASTA index file
-      pattern: "*.{fa.fai,fasta.fai}"
   - target_bed:
       type: file
       description: BED file containing target regions for variant calling
@@ -55,6 +47,14 @@ input:
       type: file
       description: Index for BED file containing target regions for variant calling
       pattern: "*.{bed.tbi}"
+  - fasta:
+      type: file
+      description: Genome reference FASTA file
+      pattern: "*.{fa,fasta}"
+  - fai:
+      type: file
+      description: Genome reference FASTA index file
+      pattern: "*.{fa.fai,fasta.fai}"
 
 output:
   - meta:

modules/manta/tumoronly/main.nf

@@ -8,11 +8,9 @@ process MANTA_TUMORONLY {
         'quay.io/biocontainers/manta:1.6.0--h9ee0642_1' }"
 
     input:
-    tuple val(meta), path(input), path(input_index)
+    tuple val(meta), path(input), path(input_index), path(target_bed), path(target_bed_tbi)
     path fasta
     path fai
-    path target_bed
-    path target_bed_tbi
 
     output:
     tuple val(meta), path("*candidate_small_indels.vcf.gz")    , emit: candidate_small_indels_vcf

modules/manta/tumoronly/meta.yml

@@ -31,14 +31,6 @@ input:
       type: file
       description: BAM/CRAM/SAM index file
       pattern: "*.{bai,crai,sai}"
-  - fasta:
-      type: file
-      description: Genome reference FASTA file
-      pattern: "*.{fa,fasta}"
-  - fai:
-      type: file
-      description: Genome reference FASTA index file
-      pattern: "*.{fa.fai,fasta.fai}"
   - target_bed:
       type: file
       description: BED file containing target regions for variant calling
@@ -47,6 +39,14 @@ input:
       type: file
       description: Index for BED file containing target regions for variant calling
       pattern: "*.{bed.tbi}"
+  - fasta:
+      type: file
+      description: Genome reference FASTA file
+      pattern: "*.{fa,fasta}"
+  - fai:
+      type: file
+      description: Genome reference FASTA index file
+      pattern: "*.{fa.fai,fasta.fai}"
 
 output:
   - meta:

tests/modules/manta/germline/main.nf

@@ -7,30 +7,28 @@ include { MANTA_GERMLINE } from '../../../../modules/manta/germline/main.nf'
 workflow test_manta_germline {
     input = [
         [ id:'test'], // meta map
-        [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_cram'], checkIfExists: true)],
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_cram'], checkIfExists: true),
-        [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_cram_crai'], checkIfExists: true)]
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_cram_crai'], 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)
-    bed     = [[],[]]
 
-    MANTA_GERMLINE ( input, fasta, fai, bed )
+    MANTA_GERMLINE ( input, fasta, fai )
 }
 
 workflow test_manta_germline_target_bed {
     input = [
         [ id:'test'], // meta map
-        [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_cram'], checkIfExists: true)],
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_cram'], checkIfExists: true),
-        [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_cram_crai'], checkIfExists: true)]
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_cram_crai'], checkIfExists: true),
+        file(params.test_data['homo_sapiens']['genome']['genome_bed_gz'], checkIfExists: true),
+        file(params.test_data['homo_sapiens']['genome']['genome_bed_gz_tbi'], 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)
-    bed     = [
-        file(params.test_data['homo_sapiens']['genome']['genome_bed_gz'], checkIfExists: true),
-        file(params.test_data['homo_sapiens']['genome']['genome_bed_gz_tbi'], checkIfExists: true),
-    ]
 
-    MANTA_GERMLINE ( input, fasta, fai, bed )
+    MANTA_GERMLINE ( input, fasta, fai )
 }
 
 workflow test_manta_germline_target_bed_jointcalling {
@@ -39,14 +37,12 @@ workflow test_manta_germline_target_bed_jointcalling {
         [file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_cram'], checkIfExists: true),
         file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_sorted_cram'], checkIfExists: true)],
         [file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_cram_crai'], checkIfExists: true),
-        file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_sorted_cram_crai'], checkIfExists: true),]
+        file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_sorted_cram_crai'], checkIfExists: true),],
+        file(params.test_data['homo_sapiens']['genome']['genome_bed_gz'], checkIfExists: true),
+        file(params.test_data['homo_sapiens']['genome']['genome_bed_gz_tbi'], 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)
-    bed     = [
-        file(params.test_data['homo_sapiens']['genome']['genome_bed_gz'], checkIfExists: true),
-        file(params.test_data['homo_sapiens']['genome']['genome_bed_gz_tbi'], checkIfExists: true),
-    ]
 
-    MANTA_GERMLINE ( input, fasta, fai, bed )
+    MANTA_GERMLINE ( input, fasta, fai )
 }

tests/modules/manta/somatic/main.nf

@@ -11,13 +11,30 @@ workflow test_manta_somatic {
         file(params.test_data['homo_sapiens']['illumina']['test_paired_end_recalibrated_sorted_cram'], checkIfExists: true),
         file(params.test_data['homo_sapiens']['illumina']['test_paired_end_recalibrated_sorted_cram_crai'], checkIfExists: true),
         file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_recalibrated_sorted_cram'], checkIfExists: true),
-        file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_recalibrated_sorted_cram_crai'], checkIfExists: true)
+        file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_recalibrated_sorted_cram_crai'], checkIfExists: true),
+        [], []
     ]
 
-    fasta   = file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
+    fasta   = file(params.test_data['homo_sapiens']['genome']['genome_21_fasta'], checkIfExists: true)
-    fai     = file(params.test_data['homo_sapiens']['genome']['genome_fasta_fai'], checkIfExists: true)
+    fai     = file(params.test_data['homo_sapiens']['genome']['genome_21_fasta_fai'], checkIfExists: true)
-    bed     = file(params.test_data['homo_sapiens']['genome']['genome_bed_gz'], checkIfExists: true)
-    bed_tbi = file(params.test_data['homo_sapiens']['genome']['genome_bed_gz_tbi'], checkIfExists: true)
 
-    MANTA_SOMATIC ( input, fasta, fai, bed, bed_tbi )
+    MANTA_SOMATIC ( input, fasta, fai )
+}
+
+workflow test_manta_somatic_target_bed {
+
+    input = [
+        [ id:'test', single_end:false ], // meta map
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_recalibrated_sorted_cram'], checkIfExists: true),
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_recalibrated_sorted_cram_crai'], checkIfExists: true),
+        file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_recalibrated_sorted_cram'], checkIfExists: true),
+        file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_recalibrated_sorted_cram_crai'], checkIfExists: true),
+        file(params.test_data['homo_sapiens']['genome']['genome_21_multi_interval_bed_gz'], checkIfExists: true),
+        file(params.test_data['homo_sapiens']['genome']['genome_21_multi_interval_bed_gz_tbi'], 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)
+
+    MANTA_SOMATIC ( input, fasta, fai )
 }

tests/modules/manta/tumoronly/main.nf

@@ -8,28 +8,27 @@ workflow test_manta_tumoronly {
     input = [
         [ id:'test'], // meta map
         file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_recalibrated_sorted_cram'], checkIfExists: true),
-        file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_recalibrated_sorted_cram_crai'], checkIfExists: true)
+        file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_recalibrated_sorted_cram_crai'], checkIfExists: true),
+        [], []
     ]
 
-    fasta   = file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
+    fasta   = file(params.test_data['homo_sapiens']['genome']['genome_21_fasta'], checkIfExists: true)
-    fai     = file(params.test_data['homo_sapiens']['genome']['genome_fasta_fai'], checkIfExists: true)
+    fai     = file(params.test_data['homo_sapiens']['genome']['genome_21_fasta_fai'], checkIfExists: true)
-    bed     = []
-    bed_tbi = []
 
-    MANTA_TUMORONLY ( input, fasta, fai, bed, bed_tbi )
+    MANTA_TUMORONLY ( input, fasta, fai )
 }
 
 workflow test_manta_tumoronly_target_bed {
     input = [
         [ id:'test'], // meta map
         file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_recalibrated_sorted_cram'], checkIfExists: true),
-        file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_recalibrated_sorted_cram_crai'], checkIfExists: true)
+        file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_recalibrated_sorted_cram_crai'], checkIfExists: true),
+        file(params.test_data['homo_sapiens']['genome']['genome_21_multi_interval_bed_gz'], checkIfExists: true),
+        file(params.test_data['homo_sapiens']['genome']['genome_21_multi_interval_bed_gz_tbi'], checkIfExists: true)
     ]
 
-    fasta   = file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
+    fasta   = file(params.test_data['homo_sapiens']['genome']['genome_21_fasta'], checkIfExists: true)
-    fai     = file(params.test_data['homo_sapiens']['genome']['genome_fasta_fai'], checkIfExists: true)
+    fai     = file(params.test_data['homo_sapiens']['genome']['genome_21_fasta_fai'], checkIfExists: true)
-    bed     = file(params.test_data['homo_sapiens']['genome']['genome_bed_gz'], checkIfExists: true)
-    bed_tbi = file(params.test_data['homo_sapiens']['genome']['genome_bed_gz_tbi'], checkIfExists: true)
 
-    MANTA_TUMORONLY ( input, fasta, fai, bed, bed_tbi )
+    MANTA_TUMORONLY ( input, fasta, fai )
 }