Merge pull request #74 from UMCUGenetics/release/v1.11.0

v1.11.0
UMCUGenetics · Jun 5, 2023 · a2ca732 · a2ca732
2 parents b81631f + 2a0cc44
commit a2ca732
Show file tree

Hide file tree

Showing 5 changed files with 119 additions and 86 deletions.
diff --git a/CustomModules b/CustomModules
diff --git a/NextflowModules b/NextflowModules
diff --git a/README.md b/README.md
@@ -1,4 +1,4 @@
-# DxNextflowWES [![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.4551799.svg)](https://doi.org/10.5281/zenodo.4551799)
+# DxNextflowWES [![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.4551798.svg)](https://doi.org/10.5281/zenodo.4551798)
 
 Genome Diagnostics Nextflow WES workflow
 

diff --git a/WES.config b/WES.config
@@ -6,6 +6,8 @@ params {
 
     gatk_rtc_options = '--known /hpc/diaggen/data/databases/GATK_resource_bundle/b37/1000G_phase1.indels.b37.vcf.gz --known /hpc/diaggen/data/databases/GATK_resource_bundle/b37/Mills_and_1000G_gold_standard.indels.b37.vcf.gz'
 
+    gatk_ir_options = '--maxReadsInMemory 150000'
+
     gatk_hc_interval_list = 'Tracks/ENSEMBL_UCSC_merged_collapsed_sorted_v3_CREv2_SSv7_100bpflank.interval_list'
     gatk_hc_interval_list_snparray = 'SNPArray/InfiniumOmni5-4v1-2.interval_list'
     gatk_hc_options = '-stand_call_conf 10 --dbsnp /hpc/diaggen/data/databases/GATK_resource_bundle/b37/dbsnp_137.b37.vcf.gz'
@@ -30,10 +32,12 @@ params {
     picard_target = 'Tracks/ENSEMBL_UCSC_merged_collapsed_sorted_v3_20bpflank.list'
 
     // following resources are inside the VerifyBamID2 container (quay.io/biocontainers/verifybamid2:2.0.1--h32f71e1_2).
-    contamination_path_prefix = '/usr/local/share/verifybamid2-2.0.1-2/resource/exome/1000g.phase3.10k.b37.exome.vcf.gz.dat'
-    contamination_sites_ud = '/usr/local/share/verifybamid2-2.0.1-2/resource/exome/1000g.phase3.10k.b37.exome.vcf.gz.dat.UD'
-    contamination_sites_mu = '/usr/local/share/verifybamid2-2.0.1-2/resource/exome/1000g.phase3.10k.b37.exome.vcf.gz.dat.mu'
-    contamination_sites_bed = '/usr/local/share/verifybamid2-2.0.1-2/resource/exome/1000g.phase3.10k.b37.exome.vcf.gz.dat.bed'
+    contamination_path_prefix = "$params.dxtracks_path/VerifyBAMID/1000g.phase3.10k.b37.exome.vcf.gz.dat"
+    contamination_sites_ud = "$params.dxtracks_path/VerifyBAMID/1000g.phase3.10k.b37.exome.vcf.gz.dat.UD"
+    contamination_sites_mu = "$params.dxtracks_path/VerifyBAMID/1000g.phase3.10k.b37.exome.vcf.gz.dat.mu"
+    contamination_sites_bed = "$params.dxtracks_path/VerifyBAMID/1000g.phase3.10k.b37.exome.vcf.gz.dat.bed"
+
+    downsample_coverage = 100
 
     vcftools_path = '/hpc/local/CentOS7/cog_bioinf/vcftools-0.1.14/bin'
     plink_path = '/hpc/local/CentOS7/cog_bioinf/plink_1.9b3/'
@@ -51,7 +55,7 @@ params {
 process {
     withLabel: BAF_IGV {
         cpus = 2
-        memory = { 5.GB * task.attempt }
+        memory = { 1.GB * task.attempt }
         time = { (1.ms * vcf_file.size() / 110) * task.attempt }
 
         publishDir {
@@ -62,13 +66,13 @@ process {
 
     withLabel: BWA_0_7_17_Mem {
         cpus = 10
-        memory = { 20.GB * task.attempt }
+        memory = { 10.GB * task.attempt }
         time = { (1.ms * fastq.sum{it.size()} / 600) * task.attempt }
     }
 
     withLabel: CheckQC {
         cpus = 2
-        memory = { 5.GB * task.attempt }
+        memory = { 1.GB * task.attempt }
         time = { 5.m * task.attempt }
 
         publishDir {
@@ -79,13 +83,13 @@ process {
 
     withLabel: ClarityEpp_SampleIndications {
         cpus = 2
-        memory = { 5.GB * task.attempt }
+        memory = { 1.GB * task.attempt }
         time = { 5.m * task.attempt }
     }
 
     withLabel: CreateHSmetricsSummary {
         cpus = 2
-        memory = { 5.GB * task.attempt }
+        memory = { 1.GB * task.attempt }
         time = { 5.m * task.attempt }
 
         publishDir {
@@ -96,7 +100,7 @@ process {
 
     withLabel: ExomeDepth_CallCNV {
         cpus = 4
-        memory = { 20.GB * task.attempt }
+        memory = { 15.GB * task.attempt }
         time = { (1.ms * bam_file.size() / 1400) * task.attempt }
 
         publishDir = [
@@ -109,7 +113,7 @@ process {
 
     withLabel: ExomeDepth_FamilyIGV {
         cpus = 2
-        memory = { 5.GB * task.attempt }
+        memory = { 1.GB * task.attempt }
         time = { 5.m * task.attempt }
 
         publishDir = [
@@ -119,13 +123,13 @@ process {
 
     withLabel: ExomeDepth_GetRefset {
         cpus = 2
-        memory = { 5.GB * task.attempt }
+        memory = { 1.GB * task.attempt }
         time = { 5.m * task.attempt }
     }
 
     withLabel: ExomeDepth_SingleIGV {
         cpus = 2
-        memory = { 5.GB * task.attempt }
+        memory = { 1.GB * task.attempt }
         time = { 5.m * task.attempt }
 
         publishDir = [
@@ -135,7 +139,7 @@ process {
 
     withLabel: ExomeDepth_Summary {
         cpus = 2
-        memory = { 2.GB * task.attempt }
+        memory = { 1.GB * task.attempt }
         time = { 5.m * task.attempt }
 
         publishDir {
@@ -152,7 +156,7 @@ process {
 
     withLabel: ExonCov_SampleQC {
         cpus = 2
-        memory = { 5.GB * task.attempt }
+        memory = { 1.GB * task.attempt }
         time = { 5.m * task.attempt }
 
         publishDir {
@@ -173,10 +177,16 @@ process {
         }
     }
 
+    withLabel: Fraction {
+        cpus = 2
+        memory = { 1.GB * task.attempt }
+        time = { 1.m * task.attempt }
+    }
+
     withLabel: GATK_3_8_1_0_gf15c1c3ef_CombineVariants {
         cpus = 2
         memory = { 5.GB * task.attempt }
-        time = { (1.ms * vcf_files.sum{it.size()} / 500) * task.attempt }
+        time = { (1.ms * vcf_files.sum{it.size()} / 450) * task.attempt }
 
         publishDir {
             path = "$params.outdir"
@@ -193,19 +203,19 @@ process {
     withLabel: GATK_3_8_1_0_gf15c1c3ef_IndelRealigner {
         cpus = 2
         memory = { 5.GB * task.attempt }
-        time = { (1.ms * bam_file.size() / 6500) * task.attempt }
+        time = { (1.ms * bam_file.size() / 6000) * task.attempt }
     }
 
     withLabel: GATK_3_8_1_0_gf15c1c3ef_RealignerTargetCreator {
         cpus = 2
         memory = { 5.GB * task.attempt }
-        time = { (1.ms * bam_file.size() / 15000) * task.attempt }
+        time = { (1.ms * bam_file.size() / 14000) * task.attempt }
     }
 
     withLabel: GATK_3_8_1_0_gf15c1c3ef_SelectVariantsSample {
         cpus = 2
         memory = { 5.GB * task.attempt }
-        time = { (1.ms * vcf_file.size() / 2000) * task.attempt }
+        time = { (1.ms * vcf_file.size() / 1500) * task.attempt }
 
         publishDir {
             path = "$params.outdir/single_sample_vcf"
@@ -227,7 +237,7 @@ process {
     withLabel: GATK_3_8_1_0_gf15c1c3ef_VariantFiltrationSnpIndel {
         cpus = 2
         memory = { 5.GB * task.attempt }
-        time = { (1.ms * vcf_file.size() + 1.s) * task.attempt }
+        time = { (1.ms * vcf_file.size() + 1.m) * task.attempt }
     }
 
     withLabel: GATK_4_2_1_0_GenotypeGVCF {
@@ -239,7 +249,7 @@ process {
     withLabel: GATK_4_2_1_0_HaplotypeCallerGVCF {
         cpus = 2
         memory = { 10.GB * task.attempt }
-        time = { (1.ms * bam_file.size() / 30000) * task.attempt }
+        time = { (1.ms * bam_file.size() / 25000) * task.attempt }
     }
 
     withLabel: GATK_4_2_1_0_MergeVcfs {
@@ -255,7 +265,7 @@ process {
 
     withLabel: GetStatsFromFlagstat {
         cpus = 2
-        memory = { 2.GB * task.attempt }
+        memory = { 1.GB * task.attempt }
         time = { 5.m * task.attempt }
 
         publishDir {
@@ -266,7 +276,7 @@ process {
 
     withLabel: Kinship {
         cpus = 2
-        memory = { 2.GB * task.attempt }
+        memory = { 1.GB * task.attempt }
         time = { 5.m * task.attempt }
 
         publishDir {
@@ -275,9 +285,16 @@ process {
         }
     }
 
+    withLabel: Mosdepth_0_3_3 {
+        cpus = 4
+        memory = { 8.GB * task.attempt }
+        time = { (1.ms * bam_file.size() / 4000) * task.attempt }
+        clusterOptions = "$params.cluster_options --gres=tmpspace:10G"
+    }
+
     withLabel: MultiQC_1_10 {
         cpus = 2
-        memory = { 5.GB * task.attempt }
+        memory = { 2.GB * task.attempt }
         time = { 5.m * task.attempt }
 
         publishDir {
@@ -288,14 +305,14 @@ process {
 
     withLabel: ParseChildFromFullTrio {
         cpus = 2
-        memory = { 5.GB * task.attempt }
+        memory = { 1.GB * task.attempt }
         time = { 5.m * task.attempt }
     }
 
     withLabel: PICARD_2_22_0_IntervalListTools {
         cpus = 2
-        memory = { 8.GB * task.attempt }
-        time = { (1.ms * interval_list.size() / 125) * task.attempt }
+        memory = { 5.GB * task.attempt }
+        time = { (1.ms * interval_list.size() / 100) * task.attempt }
         clusterOptions = "$params.cluster_options --gres=tmpspace:10G"
     }
 
@@ -325,7 +342,7 @@ process {
 
     withLabel: PICARD_2_22_0_EstimateLibraryComplexity {
         cpus = 2
-        memory = { 45.GB * task.attempt }
+        memory = { 50.GB * task.attempt }
         time = { (1.ms * bam_file.size() / 3000) * task.attempt }
         clusterOptions = "$params.cluster_options --gres=tmpspace:60G"
 
@@ -337,7 +354,7 @@ process {
 
     withLabel: Sambamba_0_7_0_Flagstat {
         cpus = 2
-        memory = { 5.GB * task.attempt }
+        memory = { 1.GB * task.attempt }
         time = { (1.ms * bam_file.size() / 3000) * task.attempt }
 
         publishDir {
@@ -348,15 +365,15 @@ process {
 
     withLabel: Sambamba_0_7_0_MarkdupMerge {
         cpus = 10
-        memory = { 50.GB * task.attempt }
+        memory = { 10.GB * task.attempt }
         time = { (1.ms * bam_files.sum{it.size()} / 8500) * task.attempt }
         clusterOptions = "$params.cluster_options --gres=tmpspace:10G"
     }
 
     withLabel: Sambamba_0_7_0_Merge {
         cpus = 10
         memory = { 5.GB * task.attempt }
-        time = { (1.ms * bam_files.sum{it.size()} / 8500) * task.attempt }
+        time = { (1.ms * bam_files.sum{it.size()} / 8000) * task.attempt }
         clusterOptions = "$params.cluster_options --gres=tmpspace:10G"
 
         publishDir {
@@ -365,6 +382,13 @@ process {
         }
     }
 
+    withLabel: Sambamba_0_7_0_ViewSubsample {
+        cpus = 10
+        memory = { 5.GB * task.attempt }
+        time = { (1.ms * bam_file.size() / 8500) * task.attempt }
+        clusterOptions = "$params.cluster_options --gres=tmpspace:10G"
+    }
+
     withLabel: Sambamba_0_7_0_ViewUnmapped {
         cpus = 2
         memory = { 1.GB * task.attempt }
@@ -384,16 +408,16 @@ process {
         }
     }
 
-    withLabel: TrendAnalysisTool {
+    withLabel: TrendAnalysis {
         cpus = 2
-        memory = { 5.GB * task.attempt }
+        memory = { 1.GB * task.attempt }
         time = { 5.m * task.attempt }
     }
 
     withLabel: UPD_IGV {
         cpus = 2
-        memory = { 5.GB * task.attempt }
-        time = { (1.ms * vcf_files.sum{it.size()} / 300) * task.attempt }
+        memory = { 1.GB * task.attempt }
+        time = { (1.ms * vcf_files.sum{it.size()} / 200) * task.attempt }
 
         publishDir = [
             [path: "$params.outdir/upd/", mode: 'link', pattern: '*.igv']
@@ -403,7 +427,7 @@ process {
     withLabel: VERIFYBAMID_2_0_1_h32f71e1_2_VerifyBamID2 {
         cpus = 6
         memory = { 5.GB * task.attempt }
-        time = { (1.ms * bam.size() / 4000) * task.attempt }
+        time = { 20.m * task.attempt }
 
         publishDir {
             path = "$params.outdir/QC/VerifyBamID2"
@@ -413,7 +437,7 @@ process {
 
     withLabel: VersionLog {
         cpus = 2
-        memory = { 5.GB * task.attempt }
+        memory = { 1.GB * task.attempt }
         time = { 5.m * task.attempt }
 
         publishDir {
@@ -424,7 +448,7 @@ process {
 
     withLabel: Workflow_Export_Params {
         cpus = 2
-        memory = { 5.GB * task.attempt }
+        memory = { 1.GB * task.attempt }
         time = { 5.m * task.attempt }
 
         publishDir {
+19 −0		Mosdepth/0.3.3/Mosdepth.nf
+18 −0		Sambamba/0.7.0/ViewSubsample.nf