Merge pull request #32 from UMCUGenetics/develop

v1.4.0, Feature contamination (verifyBamID2)

.gitignore

@@ -7,3 +7,6 @@ tools
 files
 wes_out
 fingerprint_out
+
+# VSCode
+.vscode/

WES.config

@@ -23,6 +23,12 @@ params {
     picard_bait = 'Tracks/SureSelect_CREv2_elidS30409818_Covered.list'
     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'
+
     vcftools_path = '/hpc/local/CentOS7/cog_bioinf/vcftools-0.1.14/bin'
     plink_path = '/hpc/local/CentOS7/cog_bioinf/plink_1.9b3/'
     king_path = '/hpc/local/CentOS7/cog_bioinf/king_1.4/'
@@ -234,8 +240,8 @@ process {
 
     withLabel: MultiQC_1_10 {
         cpus = 2
-        memory = '1G'
-        time = '5m'
+        memory = '5G'
+        time = '10m'
 
         publishDir {
             path = "$params.outdir/QC"
@@ -293,6 +299,16 @@ process {
         }
     }
 
+    withLabel: VERIFYBAMID_2_0_1_h32f71e1_2_VerifyBamID2 {
+        cpus = 1
+        memory = '5G'
+        time = '1h'
+
+        publishDir {
+            path = "$params.outdir/QC/VerifyBamID2"
+            mode = 'copy'
+        }
+    }
 }
 
 report {

WES.nf

@@ -32,6 +32,7 @@ include EstimateLibraryComplexity as PICARD_EstimateLibraryComplexity from './Ne
 include CollectHsMetrics as PICARD_CollectHsMetrics from './NextflowModules/Picard/2.22.0/CollectHsMetrics.nf' params(genome:"$params.genome", bait:"$params.dxtracks_path/$params.picard_bait", target:"$params.dxtracks_path/$params.picard_target", optional: "METRIC_ACCUMULATION_LEVEL=null METRIC_ACCUMULATION_LEVEL=SAMPLE")
 include Flagstat as Sambamba_Flagstat from './NextflowModules/Sambamba/0.7.0/Flagstat.nf'
 include MultiQC from './NextflowModules/MultiQC/1.10/MultiQC.nf' params(optional:"--config $baseDir/assets/multiqc_config.yaml")
+include VerifyBamID2 from './NextflowModules/VerifyBamID/2.0.1--h32f71e1_2/VerifyBamID2.nf'
 
 def fastq_files = extractFastqPairFromDir(params.fastq_path)
 def analysis_id = params.outdir.split('/')[-1]
@@ -93,11 +94,14 @@ workflow {
     Sambamba_Flagstat(Sambamba_Merge.out)
     GetStatsFromFlagstat(Sambamba_Flagstat.out.collect())
 
+    VerifyBamID2(Sambamba_Merge.out.groupTuple())
+
     MultiQC(analysis_id, Channel.empty().mix(
         FastQC.out,
         PICARD_CollectMultipleMetrics.out,
         PICARD_EstimateLibraryComplexity.out,
-        PICARD_CollectHsMetrics.out
+        PICARD_CollectHsMetrics.out,
+        VerifyBamID2.out.map{sample_id, self_sm -> [self_sm]}
     ).collect())
 
     TrendAnalysisTool(

assets/multiqc_config.yaml

@@ -86,6 +86,8 @@ table_columns_visible:
         summed_median: True
         summed_mean: False
         PERCENT_DUPLICATION: True
+    VerifyBAMID:
+        FREEMIX: True
 
 table_cond_formatting_rules:
     mqc-generalstats-picard-PCT_TARGET_BASES_20X:
@@ -95,3 +97,15 @@ table_cond_formatting_rules:
             - lt: 96
         fail:
             - lt: 95
+    mqc-generalstats-verifybamid-FREEMIX:
+        pass:
+            - lt: 5
+            - eq: 5
+        fail:
+            - gt: 5
+    FREEMIX:
+        pass:
+            - lt: 5
+            - eq: 5
+        fail:
+            - gt: 5