瀏覽代碼

rtg

master
LUYAO REN 4 年之前
父節點
當前提交
3dd043f609
共有 3 個文件被更改,包括 25 次插入23 次删除
  1. +1
    -0
      inputs
  2. +19
    -19
      tasks/benchmark.wdl
  3. +5
    -4
      workflow.wdl

+ 1
- 0
inputs 查看文件

"{{ project_name }}.SMALLcluster_config": "OnDemand bcs.ps.g.xlarge img-ubuntu-vpc", "{{ project_name }}.SMALLcluster_config": "OnDemand bcs.ps.g.xlarge img-ubuntu-vpc",
"{{ project_name }}.BIGcluster_config": "OnDemand bcs.a2.7xlarge img-ubuntu-vpc", "{{ project_name }}.BIGcluster_config": "OnDemand bcs.a2.7xlarge img-ubuntu-vpc",
"{{ project_name }}.LCL6": "{{ LCL6 }}", "{{ project_name }}.LCL6": "{{ LCL6 }}",
"{{ project_name }}.sdf": "oss://pgx-reference-data/GRCh38.d1.vd1/GRCh38.d1.vd1.sdf/",
"{{ project_name }}.MENDELIANdocker": "registry-vpc.cn-shanghai.aliyuncs.com/pgx-docker-registry/vbt:v1.1", "{{ project_name }}.MENDELIANdocker": "registry-vpc.cn-shanghai.aliyuncs.com/pgx-docker-registry/vbt:v1.1",
"{{ project_name }}.LCL7": "{{ LCL7 }}", "{{ project_name }}.LCL7": "{{ LCL7 }}",
"{{ project_name }}.DIYdocker": "registry-vpc.cn-shanghai.aliyuncs.com/pgx-docker-registry/high_confidence_call_manuscript:v1.4", "{{ project_name }}.DIYdocker": "registry-vpc.cn-shanghai.aliyuncs.com/pgx-docker-registry/high_confidence_call_manuscript:v1.4",

+ 19
- 19
tasks/benchmark.wdl 查看文件

task benchmark { task benchmark {
File vcf File vcf
File benchmarking_dir File benchmarking_dir
File ref_dir
File sdf
String sample = basename(vcf,".raw.vcf.gz") String sample = basename(vcf,".raw.vcf.gz")
String fasta String fasta
String docker String docker
set -e set -e
nt=$(nproc) nt=$(nproc)
mkdir -p /cromwell_root/tmp mkdir -p /cromwell_root/tmp
cp -r ${ref_dir} /cromwell_root/tmp/
cp -r ${benchmarking_dir} /cromwell_root/tmp/ cp -r ${benchmarking_dir} /cromwell_root/tmp/

export HGREF=/cromwell_root/tmp/reference_data/GRCh38.d1.vd1.fa
ls /cromwell_root/tmp/reference_datasets_v202103/ > files.txt


gunzip ${vcf} -c > unzip.vcf gunzip ${vcf} -c > unzip.vcf
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip unzip.vcf -c > ${sample}.rtg.vcf.gz /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip unzip.vcf -c > ${sample}.rtg.vcf.gz
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf ${sample}.rtg.vcf.gz /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf ${sample}.rtg.vcf.gz
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcffilter -i ${sample}.rtg.vcf.gz -o ${sample}.rtg.SNV.vcf.gz --snps-only
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcffilter -i ${sample}.rtg.vcf.gz -o ${sample}.rtg.INDEL.vcf.gz --non-snps-only



if [[ ${sample} =~ "LCL5" ]];then if [[ ${sample} =~ "LCL5" ]];then
/opt/hap.py/bin/hap.py /cromwell_root/tmp/reference_datasets_v202103/LCL5.high.confidence.calls.vcf.gz ${sample}.rtg.vcf.gz -f /cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed --threads $nt -o ${sample} -r ${ref_dir}/${fasta}
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg -b /cromwell_root/tmp/reference_datasets_v202103/LCL5.high.confidence.calls.SNV.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_SNV -t ${sdf} --bed-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg -b /cromwell_root/tmp/reference_datasets_v202103/LCL5.high.confidence.calls.INDEL.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_INDEL -t ${sdf} --bed-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
elif [[ ${sample} =~ "LCL6" ]]; then elif [[ ${sample} =~ "LCL6" ]]; then
/opt/hap.py/bin/hap.py /cromwell_root/tmp/reference_datasets_v202103/LCL6.high.confidence.calls.vcf.gz ${sample}.rtg.vcf.gz -f /cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed --threads $nt -o ${sample} -r ${ref_dir}/${fasta}
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg -b /cromwell_root/tmp/reference_datasets_v202103/LCL6.high.confidence.calls.SNV.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_SNV -t ${sdf} --bed-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg -b /cromwell_root/tmp/reference_datasets_v202103/LCL6.high.confidence.calls.INDEL.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_INDEL -t ${sdf} --bed-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
elif [[ ${sample} =~ "LCL7" ]]; then elif [[ ${sample} =~ "LCL7" ]]; then
/opt/hap.py/bin/hap.py /cromwell_root/tmp/reference_datasets_v202103/LCL7.high.confidence.calls.vcf.gz ${sample}.rtg.vcf.gz -f /cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed --threads $nt -o ${sample} -r ${ref_dir}/${fasta}
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg -b /cromwell_root/tmp/reference_datasets_v202103/LCL7.high.confidence.calls.SNV.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_SNV -t ${sdf} --bed-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg -b /cromwell_root/tmp/reference_datasets_v202103/LCL7.high.confidence.calls.INDEL.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_INDEL -t ${sdf} --bed-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
elif [[ ${sample} =~ "LCL8" ]]; then elif [[ ${sample} =~ "LCL8" ]]; then
/opt/hap.py/bin/hap.py /cromwell_root/tmp/reference_datasets_v202103/LCL8.high.confidence.calls.vcf.gz ${sample}.rtg.vcf.gz -f /cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed --threads $nt -o ${sample} -r ${ref_dir}/${fasta}
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg -b /cromwell_root/tmp/reference_datasets_v202103/LCL8.high.confidence.calls.SNV.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_SNV -t ${sdf} --bed-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg -b /cromwell_root/tmp/reference_datasets_v202103/LCL8.high.confidence.calls.INDEL.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_INDEL -t ${sdf} --bed-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
else else
echo "only for quartet samples" echo "only for quartet samples"
fi fi


ls /cromwell_root/tmp/reference_datasets_v202103/ > files.txt

cat ${sample}_SNV/summary.txt ${sample}_SNV_precision_recall.txt
cat ${sample}_INDEL/summary.txt ${sample}_INDEL_precision_recall.txt
>>> >>>


runtime { runtime {
output { output {
File rtg_vcf = "${sample}.rtg.vcf.gz" File rtg_vcf = "${sample}.rtg.vcf.gz"
File rtg_vcf_index = "${sample}.rtg.vcf.gz.tbi" File rtg_vcf_index = "${sample}.rtg.vcf.gz.tbi"
File gzip_vcf = "${sample}.vcf.gz"
File gzip_vcf_index = "${sample}.vcf.gz.tbi"
File roc_all_csv = "${sample}.roc.all.csv.gz"
File roc_indel = "${sample}.roc.Locations.INDEL.csv.gz"
File roc_indel_pass = "${sample}.roc.Locations.INDEL.PASS.csv.gz"
File roc_snp = "${sample}.roc.Locations.SNP.csv.gz"
File roc_snp_pass = "${sample}.roc.Locations.SNP.PASS.csv.gz"
File summary = "${sample}.summary.csv"
File extended = "${sample}.extended.csv"
File metrics = "${sample}.metrics.json.gz"
File SNV_result = "${sample}_SNV_precision_recall.txt"
File Indel_result = "${sample}_INDEL_precision_recall.txt"
File file_list = "files.txt" File file_list = "files.txt"
} }
} }

+ 5
- 4
workflow.wdl 查看文件



String fasta String fasta
File ref_dir File ref_dir
File sdf


File benchmarking_dir File benchmarking_dir
File rename File rename
input: input:
vcf=LCL5, vcf=LCL5,
benchmarking_dir=benchmarking_dir, benchmarking_dir=benchmarking_dir,
ref_dir=ref_dir,
sdf=sdf,
fasta=fasta, fasta=fasta,
docker=BENCHMARKdocker, docker=BENCHMARKdocker,
cluster_config=BIGcluster_config, cluster_config=BIGcluster_config,
input: input:
vcf=LCL6, vcf=LCL6,
benchmarking_dir=benchmarking_dir, benchmarking_dir=benchmarking_dir,
ref_dir=ref_dir,
sdf=sdf,
fasta=fasta, fasta=fasta,
docker=BENCHMARKdocker, docker=BENCHMARKdocker,
cluster_config=BIGcluster_config, cluster_config=BIGcluster_config,
input: input:
vcf=LCL7, vcf=LCL7,
benchmarking_dir=benchmarking_dir, benchmarking_dir=benchmarking_dir,
ref_dir=ref_dir,
sdf=sdf,
fasta=fasta, fasta=fasta,
docker=BENCHMARKdocker, docker=BENCHMARKdocker,
cluster_config=BIGcluster_config, cluster_config=BIGcluster_config,
input: input:
vcf=LCL8, vcf=LCL8,
benchmarking_dir=benchmarking_dir, benchmarking_dir=benchmarking_dir,
ref_dir=ref_dir,
sdf=sdf,
fasta=fasta, fasta=fasta,
docker=BENCHMARKdocker, docker=BENCHMARKdocker,
cluster_config=BIGcluster_config, cluster_config=BIGcluster_config,

Loading…
取消
儲存