Fix vcf sorting for freebayes

marine-omics · Apr 9, 2024 · 7870ec5 · 7870ec5
1 parent a1259b1
commit 7870ec5
Show file tree

Hide file tree

Showing 4 changed files with 46 additions and 8 deletions.
diff --git a/conf/base.config b/conf/base.config
@@ -13,19 +13,55 @@ process {
       cpus            = { check_max( 12 * task.attempt, 'cpus' ) }
       memory          = { check_max( 4.GB * task.attempt, 'memory' ) }
   }
+
+  withName: 'fastq2ubam'{
+    cpus   = { check_max( 2    * task.attempt, 'cpus'   ) }
+    queue='normal'
+  }
+
+  withName: 'markadapters' {
+    cpus= { check_max( 12 * task.attempt, 'cpus' ) }
+  }
+
   withName: 'bwa_mem_gatk'{
     cpus = { check_max( 12 * task.attempt, 'cpus' ) }
     memory          = { check_max( 20.GB * task.attempt, 'memory' ) }
   }
+
+  withName: 'gatk_mark_duplicates'{
+    cpus = { check_max( 4 * task.attempt, 'cpus' ) }
+    ext.args='-MAX_FILE_HANDLES 1000'
+  }
+
+  withName: 'gatk_haplotype_caller'{
+    cpus = { check_max( 4 * task.attempt, 'cpus' ) }
+    memory = { check_max( 20.GB * task.attempt, 'memory' ) }
+  }
+
+  withName: 'gatk_genomicsdb_import' {
+    cpus = { check_max( 2 * task.attempt, 'cpus' ) }
+  }
+
+  withName: 'gatk_genotypegvcfs' {
+    memory = { check_max( 8.GB * task.attempt, 'memory' ) }
+  }
+
+
+  withName: 'stat'{
+    cpus = { check_max( 12 * task.attempt, 'cpus' ) }
+  }
+
   withName: 'freebayes'{
     cpus = 1
     memory          = { check_max( 20.GB * task.attempt, 'memory' ) }
+    ext.args='-m 20 -q 20 -R 0 -S 0'
     queue = 'normal'
   }
+
   withName: 'mpileup_call'{
-    cpus = { check_max( 24 * task.attempt, 'cpus' ) }
-    memory          = { check_max( 40.GB * task.attempt, 'memory' ) }
-    queue = 'normal'
+    cpus = { check_max( 36 * task.attempt, 'cpus' ) }
+    memory          = { check_max( 80.GB * task.attempt, 'memory' ) }
+    queue = 'long'
   }
 
   withName: 'freebayes_collect'{

diff --git a/modules/gatk.nf b/modules/gatk.nf
@@ -156,9 +156,10 @@ process gatk_haplotype_caller {
     script:
     def args = task.ext.args ?: ''
     def outfile = "${bam.baseName}.g.vcf.gz"
+
     """
 
-    gatk --java-options "-Xmx20G -XX:GCTimeLimit=50 -XX:GCHeapFreeLimit=10" \
+    gatk --java-options "-Xmx${task.memory.giga}G" \
         HaplotypeCaller \
         --native-pair-hmm-threads $task.cpus \
         -R $genome \
@@ -183,11 +184,11 @@ process gatk_genomicsdb_import {
     def args = task.ext.args ?: ''
 
     """
-    gatk --java-options "-Xmx4g -Xms4g" \
+    gatk --java-options "-Xmx${task.memory.giga}g -Xms${task.memory.giga}g" \
         GenomicsDBImport \
         --genomicsdb-workspace-path ${region}.gatk.db \
         --sample-name-map $samplemap \
-        --reader-threads 4 \
+        --reader-threads ${task.cpus} \
         --batch-size 50 \
         --L $region
     """
@@ -212,7 +213,7 @@ process gatk_genotypegvcfs {
     def args = task.ext.args ?: ''
 
     """
-    gatk --java-options "-Xmx8g -Xms8g" \
+    gatk --java-options "-Xmx${task.memory.giga}g -Xms${task.memory.giga}g" \
         GenotypeGVCFs \
         -R $genome \
         -O ${region}.vcf.gz \

diff --git a/nextflow.config b/nextflow.config
@@ -63,6 +63,7 @@ profiles {
 
 
     genomics {
+        singularity.runOptions='--bind /fast/tmp:/tmp'
         process.executor = 'local'
         singularity.enabled = true
         singularity.autoMounts = true 

diff --git a/shell/sort_vcf_files.sh b/shell/sort_vcf_files.sh
@@ -1 +1 @@
-ls *.vcf | awk -F "[:.-]" '{print $1,$2,$3,$4,$5}' | sort -k2,2 -k4,4n | awk '{printf("%s.%s.%s:%s-%s.vcf\n",$1,$2,$3,$4,$5)}' 
+for f in *.vcf;do f1=${f%.vcf}; f2=${f1#freebayes.}; echo $f2 | awk -F "[:-]" '{print $1,$2,$3}';done | sort -k2,2 -k4,4n | awk '{printf("freebayes.%s:%s-%s.vcf\n",$1,$2,$3)}'