test modification

defaults

@@ -17,12 +17,16 @@
   "FASTQSCREENdocker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/fastqscreen:0.12.0",
   "SENTIEON_LICENSE": "192.168.0.55:8990",
   "SENTIEONdocker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/sentieon-genomics:v2019.11.28",
+  "REPLACE_SENTIEON_DOCKER": "xxx",
+  "DEEPVARIANT_DOCKER": "xxx",
   "QUALIMAPdocker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/qualimap:2.0.0",
+  "GATK_DOCKER": "",
   "vcf_D5": "",
   "benchmark_region": "oss://pgx-result/renluyao/manuscript_v3.0/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed",
   "db_mills": "Mills_and_1000G_gold_standard.indels.hg38.vcf",
   "dbsnp": "dbsnp_146.hg38.vcf",
   "MENDELIANdocker": "registry-vpc.cn-shanghai.aliyuncs.com/pgx-docker-registry/vbt:v1.1",
   "DIYdocker": "registry-vpc.cn-shanghai.aliyuncs.com/pgx-docker-registry/high_confidence_call_manuscript:v1.4",
-  "ref_dir": "oss://pgx-reference-data/GRCh38.d1.vd1/"
-}
+  "ref_dir": "oss://pgx-reference-data/GRCh38.d1.vd1/",
+  "pl": "ILLUMINA"
+}

inputs

@@ -28,6 +28,8 @@
   "{{ project_name }}.fastq_1_D6": "{{ fastq_1_D6 }}",
   "{{ project_name }}.fastq_1_F7": "{{ fastq_1_F7 }}",
   "{{ project_name }}.SENTIEONdocker": "{{ SENTIEONdocker }}",
+  "{{ project_name }}.REPLACE_SENTIEON_DOCKER": "{{ REPLACE_SENTIEON_DOCKER }}",
+  "{{ project_name }}.DEEPVARIANT_DOCKER": "{{ DEEPVARIANT_DOCKER }}",
   "{{ project_name }}.QUALIMAPdocker": "{{ QUALIMAPdocker }}",
   "{{ project_name }}.vcf_D5": "{{ vcf_D5 }}",
   "{{ project_name }}.benchmark_region": "{{ benchmark_region }}",

tasks/BQSR.wdl

@@ -3,7 +3,6 @@ task BQSR {
     File ref_dir
     File dbsnp_dir
     File dbmills_dir
-	String SENTIEON_INSTALL_DIR
 	String fasta
 	String dbsnp
 	String db_mills
@@ -18,17 +17,40 @@ task BQSR {
 	command <<<
 		set -o pipefail
 		set -e
-		export SENTIEON_LICENSE=192.168.0.55:8990
-		nt=$(nproc)
 
-		${SENTIEON_INSTALL_DIR}/bin/sentieon driver -r ${ref_dir}/${fasta} -t $nt -i ${realigned_bam} --algo QualCal -k ${dbsnp_dir}/${dbsnp} -k ${dbmills_dir}/${db_mills} ${sample}_recal_data.table
-
-		${SENTIEON_INSTALL_DIR}/bin/sentieon driver -r ${ref_dir}/${fasta} -t $nt -i ${realigned_bam} -q ${sample}_recal_data.table --algo QualCal -k ${dbsnp_dir}/${dbsnp} -k ${dbmills_dir}/${db_mills} ${sample}_recal_data.table.post --algo ReadWriter ${sample}.sorted.deduped.realigned.recaled.bam
-
-		${SENTIEON_INSTALL_DIR}/bin/sentieon driver -t $nt --algo QualCal --plot --before ${sample}_recal_data.table --after ${sample}_recal_data.table.post ${sample}_recal_data.csv
-
-		${SENTIEON_INSTALL_DIR}/bin/sentieon plot QualCal -o ${sample}_bqsrreport.pdf ${sample}_recal_data.csv
+		# export SENTIEON_LICENSE=192.168.0.55:8990
+		# nt=$(nproc)
+		# ${SENTIEON_INSTALL_DIR}/bin/sentieon driver -r ${ref_dir}/${fasta} -t $nt -i ${realigned_bam} --algo QualCal -k ${dbsnp_dir}/${dbsnp} -k ${dbmills_dir}/${db_mills} ${sample}_recal_data.table
+		# ${SENTIEON_INSTALL_DIR}/bin/sentieon driver -r ${ref_dir}/${fasta} -t $nt -i ${realigned_bam} -q ${sample}_recal_data.table --algo QualCal -k ${dbsnp_dir}/${dbsnp} -k ${dbmills_dir}/${db_mills} ${sample}_recal_data.table.post --algo ReadWriter ${sample}.sorted.deduped.realigned.recaled.bam
+		# ${SENTIEON_INSTALL_DIR}/bin/sentieon driver -t $nt --algo QualCal --plot --before ${sample}_recal_data.table --after ${sample}_recal_data.table.post ${sample}_recal_data.csv
+		# ${SENTIEON_INSTALL_DIR}/bin/sentieon plot QualCal -o ${sample}_bqsrreport.pdf ${sample}_recal_data.csv
 
+		gatk BaseRecalibrator \
+				-R ${ref_dir}/${fasta} \ 
+				-I ${realigned_bam} \
+				--known-sites ${dbsnp_dir}/${dbsnp}	\
+				--known-sites ${dbmills_dir}/${db_mills} \
+				-O ${sample}_recal_data.table
+		
+		gatk ApplyBQSR \
+				-R ${ref_dir}/${fasta} \
+				-I ${realigned_bam} \
+				-bqsr ${sample}_recal_data.table \
+				-O ${sample}.sorted.deduped.realigned.recaled.bam
+		
+		samtools index -@ $(nproc) -o  ${sample}.sorted.deduped.realigned.recaled.bam.bai  ${sample}.sorted.deduped.realigned.recaled.bam
+		
+		# gatk BaseRecalibrator \
+		# 		-R ${ref_dir}/${fasta} \ 
+		# 		-I ${sample}.sorted.deduped.realigned.recaled.bam \
+		# 		--known-sites ${dbsnp_dir}/${dbsnp}	\
+		# 		--known-sites ${dbmills_dir}/${db_mills} \
+		# 		-O ${sample}_recal_data.table.post
+		
+		# gatk AnalyzeCovariates \
+		# 		-before ${sample}_recal_data.table \
+		# 		-after ${sample}_recal_data.table.post \
+		# 		-csv ${sample}_recal_data.csv
 	>>>
 	runtime {
 		docker:docker
@@ -38,11 +60,11 @@ task BQSR {
 	}
 
 	output {
-		File recal_table = "${sample}_recal_data.table"
-		File recal_post = "${sample}_recal_data.table.post"
+		# File recal_table = "${sample}_recal_data.table"
+		# File recal_post = "${sample}_recal_data.table.post"
 		File recaled_bam = "${sample}.sorted.deduped.realigned.recaled.bam"
 		File recaled_bam_index = "${sample}.sorted.deduped.realigned.recaled.bam.bai"
-		File recal_csv = "${sample}_recal_data.csv"
-		File bqsrreport_pdf = "${sample}_bqsrreport.pdf"
+		# File recal_csv = "${sample}_recal_data.csv"
+		# File bqsrreport_pdf = "${sample}_bqsrreport.pdf"
 	}
 }

tasks/Dedup.wdl

@@ -1,6 +1,4 @@
 task Dedup {
-
-	String SENTIEON_INSTALL_DIR
 	File sorted_bam
 	File sorted_bam_index
 	String sample = basename(sorted_bam,".sorted.bam")
@@ -12,10 +10,17 @@ task Dedup {
 	command <<<
 		set -o pipefail
 		set -e
-		export SENTIEON_LICENSE=192.168.0.55:8990
+
 		nt=$(nproc)
-		${SENTIEON_INSTALL_DIR}/bin/sentieon driver -t $nt -i ${sorted_bam} --algo LocusCollector --fun score_info ${sample}_score.txt
-		${SENTIEON_INSTALL_DIR}/bin/sentieon driver -t $nt -i ${sorted_bam} --algo Dedup --rmdup --score_info ${sample}_score.txt --metrics ${sample}_dedup_metrics.txt ${sample}.sorted.deduped.bam	
+		# ${SENTIEON_INSTALL_DIR}/bin/sentieon driver -t $nt -i ${sorted_bam} --algo LocusCollector --fun score_info ${sample}_score.txt
+		# ${SENTIEON_INSTALL_DIR}/bin/sentieon driver -t $nt -i ${sorted_bam} --algo Dedup --rmdup --score_info ${sample}_score.txt --metrics ${sample}_dedup_metrics.txt ${sample}.sorted.deduped.bam	
+		java -jar picard.jar  MarkDuplicates \
+					-I ${sorted_bam}  \
+					-O ${sample}.sorted.deduped.bam 
+					-M ${sample}_dedup_metrics.txt \
+					--REMOVE_DUPLICATES
+					
+		samtools index -@ $(nproc) -o  ${sample}.sorted.deduped.bam.bai  ${sample}.sorted.deduped.bam
 	>>>
 	runtime {
 		docker:docker
@@ -25,8 +30,8 @@ task Dedup {
 	}
 
 	output {
-		File score = "${sample}_score.txt"
-		File dedup_metrics = "${sample}_dedup_metrics.txt"
+		# File score = "${sample}_score.txt"
+		# File dedup_metrics = "${sample}_dedup_metrics.txt"
 		File Dedup_bam = "${sample}.sorted.deduped.bam"
 		File Dedup_bam_index = "${sample}.sorted.deduped.bam.bai"
 	}

tasks/Haplotyper.wdl

@@ -1,7 +1,5 @@
 task Haplotyper {
-	
     File ref_dir
-	String SENTIEON_INSTALL_DIR
 	String fasta
 	File recaled_bam
 	File recaled_bam_index
@@ -13,9 +11,18 @@ task Haplotyper {
 command <<<
 		set -o pipefail
 		set -e
-		export SENTIEON_LICENSE=192.168.0.55:8990
-		nt=$(nproc)	
-		${SENTIEON_INSTALL_DIR}/bin/sentieon driver -r ${ref_dir}/${fasta} -t $nt -i ${recaled_bam} --algo Haplotyper ${sample}_hc.vcf
+		# export SENTIEON_LICENSE=192.168.0.55:8990
+		# nt=$(nproc)	
+		# ${SENTIEON_INSTALL_DIR}/bin/sentieon driver -r ${ref_dir}/${fasta} -t $nt -i ${recaled_bam} --algo Haplotyper ${sample}_hc.vcf
+
+		/opt/deepvariant/bin/run_deepvariant \
+			--model_type=WGS \
+			--ref=${ref_dir}/${fasta} \
+			--reads=${recaled_bam} \
+			--output_vcf=${sample}_hc.vcf \
+			--num_shards=$(nproc) 
+		
+		gatk IndexFeatureFile -I ${sample}_hc.vcf -O ${sample}_hc.vcf.idx
 	>>>
 	
 	runtime {

tasks/deduped_Metrics.wdl

@@ -14,9 +14,43 @@ task deduped_Metrics {
 	command <<<
 		set -o pipefail
 		set -e
-		export SENTIEON_LICENSE=192.168.0.55:8990
-		nt=$(nproc)
-		${SENTIEON_INSTALL_DIR}/bin/sentieon driver -r ${ref_dir}/${fasta} -t $nt -i ${Dedup_bam} --algo CoverageMetrics --omit_base_output ${sample}_deduped_coverage_metrics --algo MeanQualityByCycle ${sample}_deduped_mq_metrics.txt --algo QualDistribution ${sample}_deduped_qd_metrics.txt --algo GCBias --summary ${sample}_deduped_gc_summary.txt ${sample}_deduped_gc_metrics.txt --algo AlignmentStat ${sample}_deduped_aln_metrics.txt --algo InsertSizeMetricAlgo ${sample}_deduped_is_metrics.txt --algo QualityYield ${sample}_deduped_QualityYield.txt --algo WgsMetricsAlgo ${sample}_deduped_WgsMetricsAlgo.txt
+		# export SENTIEON_LICENSE=192.168.0.55:8990
+		# nt=$(nproc)
+		# ${SENTIEON_INSTALL_DIR}/bin/sentieon driver /
+		# -r ${ref_dir}/${fasta} -t $nt -i ${Dedup_bam} /
+		# --algo CoverageMetrics --omit_base_output ${sample}_deduped_coverage_metrics  /
+		# --algo MeanQualityByCycle ${sample}_deduped_mq_metrics.txt /
+		# --algo QualDistribution ${sample}_deduped_qd_metrics.txt /
+		# --algo GCBias --summary ${sample}_deduped_gc_summary.txt ${sample}_deduped_gc_metrics.txt /
+		# --algo AlignmentStat ${sample}_deduped_aln_metrics.txt /
+		# --algo InsertSizeMetricAlgo ${sample}_deduped_is_metrics.txt /
+		# --algo QualityYield ${sample}_deduped_QualityYield.txt /
+		# --algo WgsMetricsAlgo ${sample}_deduped_WgsMetricsAlgo.txt
+		
+		# alignment stats
+		java -jar picard.jar CollectAlignmentSummaryMetrics /
+			-I ${Dedup_bam} /
+			-O ${sample}_deduped_aln_metrics.txt /
+			-R ${ref_dir}/${fasta} /
+			--VALIDATION_STRINGENCY LENIENT
+
+		# insert size stats
+		java -jar picard.jar CollectInsertSizeMetrics /
+			-I ${Dedup_bam} /
+			-O ${sample}_deduped_is_metrics.txt /
+			-H ${sample}_deduped_is_metrics.pdf
+
+		# quality yield stats
+		java -jar picard.jar CollectQualityYieldMetrics /
+			-I ${Dedup_bam} /
+			-O ${sample}_deduped_QualityYield.txt
+
+		java -jar picard.jar CollectWgsMetrics 
+			-I ${Dedup_bam} /
+			-O ${sample}_deduped_WgsMetricsAlgo.txt /
+			-R ${ref_dir}/${fasta} /
+			--VALIDATION_STRINGENCY LENIENT
+
 	>>>
 
 	runtime {
@@ -27,15 +61,15 @@ task deduped_Metrics {
 	}
 
 	output {
-		File deduped_coverage_metrics_sample_summary = "${sample}_deduped_coverage_metrics.sample_summary"
-		File deduped_coverage_metrics_sample_statistics = "${sample}_deduped_coverage_metrics.sample_statistics"
-		File deduped_coverage_metrics_sample_interval_statistics = "${sample}_deduped_coverage_metrics.sample_interval_statistics"
-		File deduped_coverage_metrics_sample_cumulative_coverage_proportions = "${sample}_deduped_coverage_metrics.sample_cumulative_coverage_proportions"
-		File deduped_coverage_metrics_sample_cumulative_coverage_counts = "${sample}_deduped_coverage_metrics.sample_cumulative_coverage_counts"
-		File deduped_mean_quality = "${sample}_deduped_mq_metrics.txt"
-		File deduped_qd_metrics = "${sample}_deduped_qd_metrics.txt"
-		File deduped_gc_summary = "${sample}_deduped_gc_summary.txt"
-		File deduped_gc_metrics = "${sample}_deduped_gc_metrics.txt"
+		# File deduped_coverage_metrics_sample_summary = "${sample}_deduped_coverage_metrics.sample_summary"
+		# File deduped_coverage_metrics_sample_statistics = "${sample}_deduped_coverage_metrics.sample_statistics"
+		# File deduped_coverage_metrics_sample_interval_statistics = "${sample}_deduped_coverage_metrics.sample_interval_statistics"
+		# File deduped_coverage_metrics_sample_cumulative_coverage_proportions = "${sample}_deduped_coverage_metrics.sample_cumulative_coverage_proportions"
+		# File deduped_coverage_metrics_sample_cumulative_coverage_counts = "${sample}_deduped_coverage_metrics.sample_cumulative_coverage_counts"
+		# File deduped_mean_quality = "${sample}_deduped_mq_metrics.txt"
+		# File deduped_qd_metrics = "${sample}_deduped_qd_metrics.txt"
+		# File deduped_gc_summary = "${sample}_deduped_gc_summary.txt"
+		# File deduped_gc_metrics = "${sample}_deduped_gc_metrics.txt"
 		File dedeuped_aln_metrics = "${sample}_deduped_aln_metrics.txt"
 		File deduped_is_metrics = "${sample}_deduped_is_metrics.txt"
 		File deduped_QualityYield = "${sample}_deduped_QualityYield.txt"

tasks/mapping.wdl

@@ -5,8 +5,6 @@ task mapping {
 	File fastq_1
 	File fastq_2
 
-	String SENTIEON_INSTALL_DIR
-	String SENTIEON_LICENSE
 	String group
 	String sample
 	String project
@@ -19,9 +17,16 @@ task mapping {
 	command <<<
 		set -o pipefail
 		set -e	
-		export SENTIEON_LICENSE=${SENTIEON_LICENSE}
-		nt=$(nproc)
-		${SENTIEON_INSTALL_DIR}/bin/bwa mem -M -R "@RG\tID:${group}\tSM:${sample}\tPL:${pl}" -t $nt -K 10000000 ${ref_dir}/${fasta} ${fastq_1} ${fastq_2} | ${SENTIEON_INSTALL_DIR}/bin/sentieon util sort -o ${user_define_name}_${project}_${sample}.sorted.bam -t $nt --sam2bam -i -
+		# export SENTIEON_LICENSE=${SENTIEON_LICENSE}
+		# nt=$(nproc)
+		# ${SENTIEON_INSTALL_DIR}/bin/bwa mem -M -R "@RG\tID:${group}\tSM:${sample}\tPL:${pl}" -t $nt -K 10000000 ${ref_dir}/${fasta} ${fastq_1} ${fastq_2} | ${SENTIEON_INSTALL_DIR}/bin/sentieon util sort -o ${user_define_name}_${project}_${sample}.sorted.bam -t $nt --sam2bam -i -
+		bwa mem -M -R "@RG\tID:${group}\tSM:${sample}\tPL:${pl}" -t $(nproc) -K 10000000 ${ref_dir}/${fasta} ${fastq_1} ${fastq_2} \
+			| samtools view -bS -@ $(nproc) - \
+			| samtools sort -@ $(nproc) -o ${user_define_name}_${project}_${sample}.sorted.bam -
+		
+		samtools index -@ $(nproc) \
+						-o ${user_define_name}_${project}_${sample}.sorted.bam.bai  \
+						${user_define_name}_${project}_${sample}.sorted.bam
 	>>>
 
 	runtime {

tasks/sentieon.wdl

@@ -5,30 +5,36 @@ task sentieon {
 	File is_metrics
 
 	String sample = basename(quality_yield,"_deduped_QualityYield.txt")
-	String docker
 	String cluster_config
 	String disk_size
 
 	command <<<
-		set -o pipefail
+		set -o pipefail	
 		set -e
 
-		cat ${quality_yield} | sed -n '2,2p' > quality_yield.header
-		cat ${quality_yield} | sed -n '3,3p' > ${sample}.quality_yield
+		# cat ${quality_yield} | sed -n '2,2p' > quality_yield.header
+		# cat ${quality_yield} | sed -n '3,3p' > ${sample}.quality_yield
+		cat ${quality_yield} | sed -n '7,7p' > quality_yield.header
+		cat ${quality_yield} | sed -n '8,8p' > ${sample}.quality_yield
 
-		cat ${wgs_metrics_algo} | sed -n '2,2p' > wgs_metrics_algo.header
-		cat ${wgs_metrics_algo} | sed -n '3,3p' > ${sample}.wgs_metrics_algo
+		# cat ${wgs_metrics_algo} | sed -n '2,2p' > wgs_metrics_algo.header
+		# cat ${wgs_metrics_algo} | sed -n '3,3p' > ${sample}.wgs_metrics_algo
+		cat ${wgs_metrics_algo} | sed -n '7,7p' > wgs_metrics_algo.header
+		cat ${wgs_metrics_algo} | sed -n '8,8p' > ${sample}.wgs_metrics_algo
 
-		cat ${aln_metrics} | sed -n '2,2p'  > aln_metrics.header
-		cat ${aln_metrics} | sed -n '5,5p'  > ${sample}.aln_metrics
+		# cat ${aln_metrics} | sed -n '2,2p'  > aln_metrics.header
+		# cat ${aln_metrics} | sed -n '5,5p'  > ${sample}.aln_metrics
+		cat ${aln_metrics} | sed -n '7,7p'  > aln_metrics.header
+		cat ${aln_metrics} | sed -n '10,10p'  > ${sample}.aln_metrics
 
-		cat ${is_metrics} | sed -n '2,2p' > is_metrics.header
-		cat ${is_metrics} | sed -n '3,3p' > ${sample}.is_metrics
+		# cat ${is_metrics} | sed -n '2,2p' > is_metrics.header
+		# cat ${is_metrics} | sed -n '3,3p' > ${sample}.is_metrics
+		cat ${is_metrics} | sed -n '7,7p' > is_metrics.header
+		cat ${is_metrics} | sed -n '8,8p' > ${sample}.is_metrics
 
 	>>>
 
 	runtime {
-		docker:docker
 		cluster:cluster_config
 		systemDisk:"cloud_ssd 40"
 		dataDisk:"cloud_ssd " + disk_size + " /cromwell_root/"