finalize merge_gvcfs.nf

seqeralabs · Oct 21, 2020 · 0b6c311 · 0b6c311
1 parent 58eabe1
commit 0b6c311
Show file tree

Hide file tree

Showing 4 changed files with 28 additions and 28 deletions.
diff --git a/modules/gatk/haplotype_caller/haplotype_caller.nf b/modules/gatk/haplotype_caller/haplotype_caller.nf
@@ -3,7 +3,7 @@
 //===========================
 
 params.contamination
-params.make_gvcf
+params.make_gvcf = false
 params.make_bamout
 params.hc_scatter
 params.gcs_project_for_requester_pays = false
@@ -17,7 +17,7 @@ process GATK_HAPLOTYPE_CALLER {
 //---------------------------
 //  directives
 //---------------------------
-
+    container = "broadinstitute/genomes-in-the-cloud:2.3.1-1500064817"
 
 //---------------------------
     input:

diff --git a/modules/gatk/merge_gvcfs/merge_gvcfs.nf b/modules/gatk/merge_gvcfs/merge_gvcfs.nf
@@ -2,11 +2,8 @@
 // Module parameters
 //===========================
 
-params.contamination
-params.output_filename
-params.gatk_path
-params.machine_mem_gb
-params.command_mem_gb
+params.make_gvcf = false
+params.memory = '10'
 
 //===========================
 // Process definition
@@ -16,11 +13,10 @@ process GATK_MERGE_GVCFS {
 
 
 //---------------------------
-// directives
+//  directives
 //---------------------------
-    container
-    memory
-    disk
+    container = "broadinstitute/genomes-in-the-cloud:2.3.1-1500064817"
+    memory "${params.memory}GB"
     // NOTE: WDL pre-emptible is not applicable. See https://cromwell.readthedocs.io/en/stable/RuntimeAttributes/#preemptible
     // NOTE: Instead of WDL preemptive, we can rely on the excutor-independent `errorStrategy` and `maxRetries` directives in NXF
     errorStrategy
@@ -29,25 +25,28 @@ process GATK_MERGE_GVCFS {
 //---------------------------
     input:
 //---------------------------
-    path(input_vcfs)
-    path(input_vcfs_indexes)
+    tuple path(input_vcf), path(input_vcf_index)
 
 
 //---------------------------
     output:
 //---------------------------
+    tuple path("${output_filename}"), path("${output_filename}.tbi")
 
 //---------------------------
     script:
 //---------------------------
+    input_str = input_vcf.reduce("") { a, b -> a + " --INPUT ${b}" }.join(' ')
+    output_suffix = params.make_gvcf ? ".g.vcf.gz" : ".vcf.gz"
+    output_filename = input_vcf.getBaseName() + output_suffix
 
     """
     set -e
 
-    ${gatk_path} --java-options "-Xmx${command_mem_gb}G"  \
+    ${gatk_path} --java-options "-Xmx${$params.memory}G"  \
       MergeVcfs \
-      --INPUT ${sep = ' --INPUT ' input_vcfs} \
-      --OUTPUT ~{output_filename}
+      --INPUT ${input_str} \
+      --OUTPUT ${output_filename}
     
-"""
+    """
 }
diff --git a/modules/samtools/cram_to_bam/cram_to_bam.nf b/modules/samtools/cram_to_bam/cram_to_bam.nf
@@ -1,7 +1,7 @@
 //===========================
 // Module parameters
 //===========================
-
+params.samtools_path
 
 //===========================
 // Process definition
@@ -12,13 +12,7 @@ process SAMTOOLS_CRAM_TO_BAM {
 //---------------------------
 // directives
 //---------------------------
-    // TODO: We could add the samtools docker image
     container = "broadinstitute/genomes-in-the-cloud:2.3.1-1500064817"
-    // TODO: Allocate these resources dynamically. See Line-135-137 of haplotypecaller_gvcf_gatk4.wdl
-    //    memory
-    //    disk
-    // NOTE: WDL pre-emptible is not applicable. See https://cromwell.readthedocs.io/en/stable/RuntimeAttributes/#preemptible
-    // NOTE: Instead of WDL preemptive, we can rely on the excutor-independent `errorStrategy` and `maxRetries` directives in NXF
 
 
 //---------------------------
@@ -28,7 +22,7 @@ process SAMTOOLS_CRAM_TO_BAM {
     path(ref_fasta_index)
     path(ref_dict)
     path(input_cram)
-    path(sample_name)
+
 
 //---------------------------
     output:
@@ -40,13 +34,15 @@ process SAMTOOLS_CRAM_TO_BAM {
 //---------------------------
     script:
 //---------------------------
+    sample_name = input_cram.getBaseName()
+
     """
     set -e
     set -o pipefall
     
-    ${samtools_path} view -h -T ${ref_fasta} ${input_cram}
-    ${samtools_path} view -b -o ${sample_name}.bam -
-    ${samtools_path} index -b ${sample_name}.bam
+    ${params.samtools_path} view -h -T ${ref_fasta} ${input_cram}
+    ${params.samtools_path} view -b -o ${sample_name}.bam -
+    ${params.samtools_path} index -b ${sample_name}.bam
     mv ${sample_name}.bam.bai ${sample_name}.bai
     
     """

diff --git a/workflows/haplotype_caller.nf b/workflows/haplotype_caller.nf
@@ -21,5 +21,10 @@ workflow {
     take:
 
     main:
+    SAMTOOLS_CRAM_TO_BAM
+    GATK_HAPLOTYPE_CALLER
+    GATK_MERGE_GVCFS
+
+    emit:
 
 }