From 7389963d5cb18f81c10dff128c510e518ee4f0f6 Mon Sep 17 00:00:00 2001
From: FriederikeHanssen <Friederike.hanssen@qbic.uni-tuebingen.de>
Date: Tue, 7 Dec 2021 15:22:24 +0100
Subject: [PATCH] Add memory stuff to all gatk4 modules (#1122)

* Add memory stuff to all gatj4 modules

* Add removed input line back in

* revert script section
---
 modules/gatk4/applybqsr/main.nf                   | 3 ++-
 modules/gatk4/baserecalibrator/main.nf            | 4 ++--
 modules/gatk4/bedtointervallist/main.nf           | 8 +++++++-
 modules/gatk4/calculatecontamination/main.nf      | 8 +++++++-
 modules/gatk4/createsequencedictionary/main.nf    | 2 +-
 modules/gatk4/createsomaticpanelofnormals/main.nf | 8 +++++++-
 modules/gatk4/estimatelibrarycomplexity/main.nf   | 2 +-
 modules/gatk4/fastqtosam/main.nf                  | 8 +++++++-
 modules/gatk4/filtermutectcalls/main.nf           | 8 +++++++-
 modules/gatk4/genomicsdbimport/main.nf            | 8 +++++++-
 modules/gatk4/genotypegvcfs/main.nf               | 8 +++++++-
 modules/gatk4/getpileupsummaries/main.nf          | 8 +++++++-
 modules/gatk4/indexfeaturefile/main.nf            | 8 +++++++-
 modules/gatk4/intervallisttools/main.nf           | 8 +++++++-
 modules/gatk4/learnreadorientationmodel/main.nf   | 8 +++++++-
 modules/gatk4/markduplicates/main.nf              | 6 +++---
 modules/gatk4/mergebamalignment/main.nf           | 8 +++++++-
 modules/gatk4/mergevcfs/main.nf                   | 8 +++++++-
 modules/gatk4/mutect2/main.nf                     | 8 +++++++-
 modules/gatk4/revertsam/main.nf                   | 8 +++++++-
 modules/gatk4/samtofastq/main.nf                  | 8 +++++++-
 modules/gatk4/splitncigarreads/main.nf            | 8 +++++++-
 modules/gatk4/variantfiltration/main.nf           | 2 +-
 23 files changed, 129 insertions(+), 26 deletions(-)

diff --git a/modules/gatk4/applybqsr/main.nf b/modules/gatk4/applybqsr/main.nf
index bd428d6c..3cc69ddf 100644
--- a/modules/gatk4/applybqsr/main.nf
+++ b/modules/gatk4/applybqsr/main.nf
@@ -22,13 +22,14 @@ process GATK4_APPLYBQSR {
     def args = task.ext.args ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
     def interval = intervals ? "-L ${intervals}" : ""
+    def avail_mem = 3
     if (!task.memory) {
         log.info '[GATK ApplyBQSR] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
     } else {
         avail_mem = task.memory.giga
     }
     """
-    gatk ApplyBQSR \\
+    gatk --java-options "-Xmx${avail_mem}g" ApplyBQSR \\
         -R $fasta \\
         -I $input \\
         --bqsr-recal-file $bqsr_table \\
diff --git a/modules/gatk4/baserecalibrator/main.nf b/modules/gatk4/baserecalibrator/main.nf
index 9b0bf286..17b37943 100644
--- a/modules/gatk4/baserecalibrator/main.nf
+++ b/modules/gatk4/baserecalibrator/main.nf
@@ -25,14 +25,14 @@ process GATK4_BASERECALIBRATOR {
     def prefix = task.ext.prefix ?: "${meta.id}"
     def intervalsCommand = intervalsBed ? "-L ${intervalsBed}" : ""
     def sitesCommand = knownSites.collect{"--known-sites ${it}"}.join(' ')
-
+    def avail_mem = 3
     if (!task.memory) {
         log.info '[GATK BaseRecalibrator] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
     } else {
         avail_mem = task.memory.giga
     }
     """
-    gatk BaseRecalibrator  \
+    gatk --java-options "-Xmx${avail_mem}g" BaseRecalibrator  \
         -R $fasta \
         -I $input \
         $sitesCommand \
diff --git a/modules/gatk4/bedtointervallist/main.nf b/modules/gatk4/bedtointervallist/main.nf
index c4538034..2f6266b9 100644
--- a/modules/gatk4/bedtointervallist/main.nf
+++ b/modules/gatk4/bedtointervallist/main.nf
@@ -18,8 +18,14 @@ process GATK4_BEDTOINTERVALLIST {
     script:
     def args = task.ext.args ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
+    def avail_mem = 3
+    if (!task.memory) {
+        log.info '[GATK BedToIntervalList] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
+    } else {
+        avail_mem = task.memory.giga
+    }
     """
-    gatk BedToIntervalList \\
+    gatk --java-options "-Xmx${avail_mem}g" BedToIntervalList \\
         -I $bed \\
         -SD $sequence_dict \\
         -O ${prefix}.interval_list \\
diff --git a/modules/gatk4/calculatecontamination/main.nf b/modules/gatk4/calculatecontamination/main.nf
index 7c112c3c..8840356a 100644
--- a/modules/gatk4/calculatecontamination/main.nf
+++ b/modules/gatk4/calculatecontamination/main.nf
@@ -21,8 +21,14 @@ process GATK4_CALCULATECONTAMINATION {
     def prefix = task.ext.prefix ?: "${meta.id}"
     def matched_command = matched ? " -matched ${matched} " : ''
     def segment_command = segmentout ? " -segments ${prefix}.segmentation.table" : ''
+    def avail_mem = 3
+    if (!task.memory) {
+        log.info '[GATK CalculateContamination] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
+    } else {
+        avail_mem = task.memory.giga
+    }
     """
-    gatk CalculateContamination \\
+    gatk --java-options "-Xmx${avail_mem}g" CalculateContamination \\
         -I $pileup \\
         $matched_command \\
         -O ${prefix}.contamination.table \\
diff --git a/modules/gatk4/createsequencedictionary/main.nf b/modules/gatk4/createsequencedictionary/main.nf
index 8d001856..e8f32106 100644
--- a/modules/gatk4/createsequencedictionary/main.nf
+++ b/modules/gatk4/createsequencedictionary/main.nf
@@ -18,7 +18,7 @@ process GATK4_CREATESEQUENCEDICTIONARY {
     def args = task.ext.args ?: ''
     def avail_mem = 6
     if (!task.memory) {
-        log.info '[GATK] Available memory not known - defaulting to 6GB. Specify process memory requirements to change this.'
+        log.info '[GATK CreateSequenceDictionary] Available memory not known - defaulting to 6GB. Specify process memory requirements to change this.'
     } else {
         avail_mem = task.memory.giga
     }
diff --git a/modules/gatk4/createsomaticpanelofnormals/main.nf b/modules/gatk4/createsomaticpanelofnormals/main.nf
index 2860e82e..ff345f75 100644
--- a/modules/gatk4/createsomaticpanelofnormals/main.nf
+++ b/modules/gatk4/createsomaticpanelofnormals/main.nf
@@ -21,8 +21,14 @@ process GATK4_CREATESOMATICPANELOFNORMALS {
     script:
     def args = task.ext.args ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
+    def avail_mem = 3
+    if (!task.memory) {
+        log.info '[GATK CreateSomaticPanelOfNormals] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
+    } else {
+        avail_mem = task.memory.giga
+    }
     """
-    gatk \\
+    gatk --java-options "-Xmx${avail_mem}g" \\
         CreateSomaticPanelOfNormals \\
         -R $fasta \\
         -V gendb://$genomicsdb \\
diff --git a/modules/gatk4/estimatelibrarycomplexity/main.nf b/modules/gatk4/estimatelibrarycomplexity/main.nf
index f636dc46..c17dba09 100644
--- a/modules/gatk4/estimatelibrarycomplexity/main.nf
+++ b/modules/gatk4/estimatelibrarycomplexity/main.nf
@@ -29,7 +29,7 @@ process GATK4_ESTIMATELIBRARYCOMPLEXITY {
         avail_mem = task.memory.giga
     }
     """
-    gatk EstimateLibraryComplexity \
+    gatk --java-options "-Xmx${avail_mem}g" EstimateLibraryComplexity \
         ${crams} \
         -O ${prefix}.metrics \
         --REFERENCE_SEQUENCE ${fasta} \
diff --git a/modules/gatk4/fastqtosam/main.nf b/modules/gatk4/fastqtosam/main.nf
index 915eb996..a55ba709 100644
--- a/modules/gatk4/fastqtosam/main.nf
+++ b/modules/gatk4/fastqtosam/main.nf
@@ -18,8 +18,14 @@ process GATK4_FASTQTOSAM {
     def args = task.ext.args ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
     def read_files = meta.single_end ? "-F1 $reads" : "-F1 ${reads[0]} -F2 ${reads[1]}"
+    def avail_mem = 3
+    if (!task.memory) {
+        log.info '[GATK FastqToSam] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
+    } else {
+        avail_mem = task.memory.giga
+    }
     """
-    gatk FastqToSam \\
+    gatk --java-options "-Xmx${avail_mem}g" FastqToSam \\
         $read_files \\
         -O ${prefix}.bam \\
         -SM $prefix \\
diff --git a/modules/gatk4/filtermutectcalls/main.nf b/modules/gatk4/filtermutectcalls/main.nf
index 02fa804f..6a1d9b3a 100644
--- a/modules/gatk4/filtermutectcalls/main.nf
+++ b/modules/gatk4/filtermutectcalls/main.nf
@@ -37,8 +37,14 @@ process GATK4_FILTERMUTECTCALLS {
     if (contaminationfile) {
         contamination_options = '--contamination-table ' + contaminationfile.join(' --contamination-table ')
     }
+    def avail_mem = 3
+    if (!task.memory) {
+        log.info '[GATK FilterMutectCalls] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
+    } else {
+        avail_mem = task.memory.giga
+    }
     """
-    gatk FilterMutectCalls \\
+    gatk --java-options "-Xmx${avail_mem}g" FilterMutectCalls \\
         -R $fasta \\
         -V $vcf \\
         $orientationbias_options \\
diff --git a/modules/gatk4/genomicsdbimport/main.nf b/modules/gatk4/genomicsdbimport/main.nf
index e794aa5a..2751173b 100644
--- a/modules/gatk4/genomicsdbimport/main.nf
+++ b/modules/gatk4/genomicsdbimport/main.nf
@@ -42,8 +42,14 @@ process GATK4_GENOMICSDBIMPORT {
         updated_db = wspace.toString()
     }
 
+    def avail_mem = 3
+    if (!task.memory) {
+        log.info '[GATK GenomicsDBImport] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
+    } else {
+        avail_mem = task.memory.giga
+    }
     """
-    gatk GenomicsDBImport \\
+    gatk --java-options "-Xmx${avail_mem}g" GenomicsDBImport \\
         $inputs_command \\
         $dir_command \\
         $intervals_command \\
diff --git a/modules/gatk4/genotypegvcfs/main.nf b/modules/gatk4/genotypegvcfs/main.nf
index f0b35447..1a772860 100644
--- a/modules/gatk4/genotypegvcfs/main.nf
+++ b/modules/gatk4/genotypegvcfs/main.nf
@@ -26,8 +26,14 @@ process GATK4_GENOTYPEGVCFS {
     def dbsnp_options    = dbsnp ? "-D ${dbsnp}" : ""
     def interval_options = intervals_bed ? "-L ${intervals_bed}" : ""
     def gvcf_options     = gvcf.name.endsWith(".vcf") || gvcf.name.endsWith(".vcf.gz") ? "$gvcf" : "gendb://$gvcf"
+    def avail_mem = 3
+    if (!task.memory) {
+        log.info '[GATK GenotypeGVCFs] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
+    } else {
+        avail_mem = task.memory.giga
+    }
     """
-    gatk \\
+    gatk --java-options "-Xmx${avail_mem}g" \\
         GenotypeGVCFs \\
         $args \\
         $interval_options \\
diff --git a/modules/gatk4/getpileupsummaries/main.nf b/modules/gatk4/getpileupsummaries/main.nf
index 99be601f..361974e8 100644
--- a/modules/gatk4/getpileupsummaries/main.nf
+++ b/modules/gatk4/getpileupsummaries/main.nf
@@ -24,8 +24,14 @@ process GATK4_GETPILEUPSUMMARIES {
 
     sitesCommand = sites ? " -L ${sites} " : " -L ${variants} "
 
+    def avail_mem = 3
+    if (!task.memory) {
+        log.info '[GATK GetPileupSummaries] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
+    } else {
+        avail_mem = task.memory.giga
+    }
     """
-    gatk GetPileupSummaries \\
+    gatk --java-options "-Xmx${avail_mem}g" GetPileupSummaries \\
         -I $bam \\
         -V $variants \\
         $sitesCommand \\
diff --git a/modules/gatk4/indexfeaturefile/main.nf b/modules/gatk4/indexfeaturefile/main.nf
index d33e030c..cc6c663e 100644
--- a/modules/gatk4/indexfeaturefile/main.nf
+++ b/modules/gatk4/indexfeaturefile/main.nf
@@ -16,8 +16,14 @@ process GATK4_INDEXFEATUREFILE {
 
     script:
     def args = task.ext.args ?: ''
+    def avail_mem = 3
+    if (!task.memory) {
+        log.info '[GATK IndexFeatureFile] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
+    } else {
+        avail_mem = task.memory.giga
+    }
     """
-    gatk \\
+    gatk --java-options "-Xmx${avail_mem}g" \\
         IndexFeatureFile \\
         $args \\
         -I $feature_file
diff --git a/modules/gatk4/intervallisttools/main.nf b/modules/gatk4/intervallisttools/main.nf
index 7e1a47f7..b813d844 100644
--- a/modules/gatk4/intervallisttools/main.nf
+++ b/modules/gatk4/intervallisttools/main.nf
@@ -17,11 +17,17 @@ process GATK4_INTERVALLISTTOOLS {
     script:
     def args = task.ext.args ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
+    def avail_mem = 3
+    if (!task.memory) {
+        log.info '[GATK IntervalListTools] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
+    } else {
+        avail_mem = task.memory.giga
+    }
     """
 
     mkdir ${prefix}_split
 
-    gatk \\
+    gatk --java-options "-Xmx${avail_mem}g" \\
         IntervalListTools \\
         -I ${interval_list} \\
         -O ${prefix}_split \\
diff --git a/modules/gatk4/learnreadorientationmodel/main.nf b/modules/gatk4/learnreadorientationmodel/main.nf
index ac021afa..0c2f09d2 100644
--- a/modules/gatk4/learnreadorientationmodel/main.nf
+++ b/modules/gatk4/learnreadorientationmodel/main.nf
@@ -19,8 +19,14 @@ process GATK4_LEARNREADORIENTATIONMODEL {
     def prefix = task.ext.prefix ?: "${meta.id}"
     def inputs_list = []
     f1r2.each() { a -> inputs_list.add(" -I " + a) }
+    def avail_mem = 3
+    if (!task.memory) {
+        log.info '[GATK LearnReadOrientationModel] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
+    } else {
+        avail_mem = task.memory.giga
+    }
     """
-    gatk \\
+    gatk --java-options "-Xmx${avail_mem}g" \\
         LearnReadOrientationModel \\
         ${inputs_list.join(' ')} \\
         -O ${prefix}.tar.gz \\
diff --git a/modules/gatk4/markduplicates/main.nf b/modules/gatk4/markduplicates/main.nf
index a109facc..8bdb2c0a 100644
--- a/modules/gatk4/markduplicates/main.nf
+++ b/modules/gatk4/markduplicates/main.nf
@@ -20,14 +20,14 @@ process GATK4_MARKDUPLICATES {
     def args = task.ext.args ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
     def bam_list = bams.collect(){ bam -> "--INPUT ".concat(bam.toString()) }.join(" ")
-    def avail_mem       = 3
+    def avail_mem = 3
     if (!task.memory) {
-        log.info '[GATK HaplotypeCaller] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
+        log.info '[GATK MarkDuplicates] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
     } else {
         avail_mem = task.memory.giga
     }
     """
-    gatk MarkDuplicates \\
+    gatk --java-options "-Xmx${avail_mem}g" MarkDuplicates \\
         $bam_list \\
         --METRICS_FILE ${prefix}.metrics \\
         --TMP_DIR . \\
diff --git a/modules/gatk4/mergebamalignment/main.nf b/modules/gatk4/mergebamalignment/main.nf
index 5e552cb2..a0f54976 100644
--- a/modules/gatk4/mergebamalignment/main.nf
+++ b/modules/gatk4/mergebamalignment/main.nf
@@ -20,8 +20,14 @@ process GATK4_MERGEBAMALIGNMENT {
     script:
     def args = task.ext.args ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
+    def avail_mem = 3
+    if (!task.memory) {
+        log.info '[GATK MergeBamAlignment] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
+    } else {
+        avail_mem = task.memory.giga
+    }
     """
-    gatk MergeBamAlignment \\
+    gatk --java-options "-Xmx${avail_mem}g" MergeBamAlignment \\
         ALIGNED=$aligned \\
         UNMAPPED=$unmapped \\
         R=$fasta \\
diff --git a/modules/gatk4/mergevcfs/main.nf b/modules/gatk4/mergevcfs/main.nf
index cd1840c3..1fcce485 100644
--- a/modules/gatk4/mergevcfs/main.nf
+++ b/modules/gatk4/mergevcfs/main.nf
@@ -26,8 +26,14 @@ process GATK4_MERGEVCFS {
         input += " I=${vcf}"
     }
     def ref = use_ref_dict ? "D=${ref_dict}" : ""
+    def avail_mem = 3
+    if (!task.memory) {
+        log.info '[GATK MergeVcfs] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
+    } else {
+        avail_mem = task.memory.giga
+    }
     """
-    gatk MergeVcfs \\
+    gatk --java-options "-Xmx${avail_mem}g" MergeVcfs \\
         $input \\
         O=${prefix}.vcf.gz \\
         $ref \\
diff --git a/modules/gatk4/mutect2/main.nf b/modules/gatk4/mutect2/main.nf
index 2cf940de..414c7705 100644
--- a/modules/gatk4/mutect2/main.nf
+++ b/modules/gatk4/mutect2/main.nf
@@ -53,8 +53,14 @@ process GATK4_MUTECT2 {
         normals_command = '-normal ' + which_norm.join( ' -normal ')
     }
 
+    def avail_mem = 3
+    if (!task.memory) {
+        log.info '[GATK Mutect2] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
+    } else {
+        avail_mem = task.memory.giga
+    }
     """
-    gatk Mutect2 \\
+    gatk --java-options "-Xmx${avail_mem}g" Mutect2 \\
         -R ${fasta} \\
         ${inputs_command} \\
         ${normals_command} \\
diff --git a/modules/gatk4/revertsam/main.nf b/modules/gatk4/revertsam/main.nf
index 638b7705..0713d7ca 100644
--- a/modules/gatk4/revertsam/main.nf
+++ b/modules/gatk4/revertsam/main.nf
@@ -17,8 +17,14 @@ process GATK4_REVERTSAM {
     script:
     def args = task.ext.args ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
+    def avail_mem = 3
+    if (!task.memory) {
+        log.info '[GATK RevertSam] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
+    } else {
+        avail_mem = task.memory.giga
+    }
     """
-    gatk RevertSam \\
+    gatk --java-options "-Xmx${avail_mem}g" RevertSam \\
         I=$bam \\
         O=${prefix}.reverted.bam \\
         $args
diff --git a/modules/gatk4/samtofastq/main.nf b/modules/gatk4/samtofastq/main.nf
index a909f540..0afb7ef3 100644
--- a/modules/gatk4/samtofastq/main.nf
+++ b/modules/gatk4/samtofastq/main.nf
@@ -18,8 +18,14 @@ process GATK4_SAMTOFASTQ {
     def args = task.ext.args ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
     def output   = meta.single_end ? "FASTQ=${prefix}.fastq.gz" : "FASTQ=${prefix}_1.fastq.gz SECOND_END_FASTQ=${prefix}_2.fastq.gz"
+    def avail_mem = 3
+    if (!task.memory) {
+        log.info '[GATK SamToFastq] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
+    } else {
+        avail_mem = task.memory.giga
+    }
     """
-    gatk SamToFastq \\
+    gatk --java-options "-Xmx${avail_mem}g" SamToFastq \\
         I=$bam \\
         $output \\
         $args
diff --git a/modules/gatk4/splitncigarreads/main.nf b/modules/gatk4/splitncigarreads/main.nf
index 65b82a35..6daed954 100644
--- a/modules/gatk4/splitncigarreads/main.nf
+++ b/modules/gatk4/splitncigarreads/main.nf
@@ -20,8 +20,14 @@ process GATK4_SPLITNCIGARREADS {
     script:
     def args = task.ext.args ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
+    def avail_mem = 3
+    if (!task.memory) {
+        log.info '[GATK SplitNCigarReads] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
+    } else {
+        avail_mem = task.memory.giga
+    }
     """
-    gatk SplitNCigarReads \\
+    gatk --java-options "-Xmx${avail_mem}g" SplitNCigarReads \\
         -R $fasta \\
         -I $bam \\
         -O ${prefix}.bam \\
diff --git a/modules/gatk4/variantfiltration/main.nf b/modules/gatk4/variantfiltration/main.nf
index 00dc2588..efe245cc 100644
--- a/modules/gatk4/variantfiltration/main.nf
+++ b/modules/gatk4/variantfiltration/main.nf
@@ -23,7 +23,7 @@ process GATK4_VARIANTFILTRATION {
     def prefix = task.ext.prefix ?: "${meta.id}"
     def avail_mem = 3
     if (!task.memory) {
-        log.info '[GATK HaplotypeCaller] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
+        log.info '[GATK VariantFiltration] Available memory not known - defaulting to 3GB. Specify process memory requirements to change this.'
     } else {
         avail_mem = task.memory.toGiga()
     }