LUYAO REN 4 anni fa
parent
commit
1f32b9846d
2 ha cambiato i file con 22 aggiunte e 6 eliminazioni
  1. +22
    -1
      tasks/benchmark.wdl
  2. +0
    -5
      tasks/merge_family.wdl

+ 22
- 1
tasks/benchmark.wdl Vedi File

@@ -23,7 +23,12 @@ task benchmark {
cat body | grep -w '^chr1\|^chr2\|^chr3\|^chr4\|^chr5\|^chr6\|^chr7\|^chr8\|^chr9\|^chr10\|^chr11\|^chr12\|^chr13\|^chr14\|^chr15\|^chr16\|^chr17\|^chr18\|^chr19\|^chr20\|^chr21\|^chr22\|^chrX' > body.filtered
cat unzip.vcf | grep '##' | grep -v 'contig' | cat - ${contig} name body.filtered > filtered.vcf

echo -e "#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tLCL5" > LCL5_name
echo -e "#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tLCL6" > LCL6_name
echo -e "#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tLCL7" > LCL7_name
echo -e "#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tLCL8" > LCL8_name


/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip filtered.vcf -c > ${project}.${sample}.rtg.vcf.gz
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf ${project}.${sample}.rtg.vcf.gz
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcffilter -i ${project}.${sample}.rtg.vcf.gz -o ${sample}.rtg.SNV.vcf.gz --snps-only
@@ -33,15 +38,31 @@ task benchmark {
if [[ ${sample} =~ "LCL5" ]];then
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -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} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -b /cromwell_root/tmp/reference_datasets_v202103/LCL5.high.confidence.calls.INDEL.vcf.gz -c ${sample}.rtg.INDEL.vcf.gz -o ${sample}_INDEL -t ${sdf} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
cat filtered.vcf | grep '##' > header
cat filtered.vcf | grep -v '#' > body
cat header LCL5_name body > LCL5.vcf
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip LCL5.vcf -c > ${project}.${sample}.rtg.vcf.gz
elif [[ ${sample} =~ "LCL6" ]]; then
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -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} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -b /cromwell_root/tmp/reference_datasets_v202103/LCL6.high.confidence.calls.INDEL.vcf.gz -c ${sample}.rtg.INDEL.vcf.gz -o ${sample}_INDEL -t ${sdf} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
cat filtered.vcf | grep '##' > header
cat filtered.vcf | grep -v '#' > body
cat header LCL6_name body > LCL6.vcf
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip LCL6.vcf -c > ${project}.${sample}.rtg.vcf.gz
elif [[ ${sample} =~ "LCL7" ]]; then
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -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} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -b /cromwell_root/tmp/reference_datasets_v202103/LCL7.high.confidence.calls.INDEL.vcf.gz -c ${sample}.rtg.INDEL.vcf.gz -o ${sample}_INDEL -t ${sdf} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
cat filtered.vcf | grep '##' > header
cat filtered.vcf | grep -v '#' > body
cat header LCL7_name body > LCL7.vcf
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip LCL7.vcf -c > ${project}.${sample}.rtg.vcf.gz
elif [[ ${sample} =~ "LCL8" ]]; then
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -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} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -b /cromwell_root/tmp/reference_datasets_v202103/LCL8.high.confidence.calls.INDEL.vcf.gz -c ${sample}.rtg.INDEL.vcf.gz -o ${sample}_INDEL -t ${sdf} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
cat filtered.vcf | grep '##' > header
cat filtered.vcf | grep -v '#' > body
cat header LCL8_name body > LCL8.vcf
/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip LCL8.vcf -c > ${project}.${sample}.rtg.vcf.gz
else
echo "only for quartet samples"
fi

+ 0
- 5
tasks/merge_family.wdl Vedi File

@@ -16,11 +16,6 @@

/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfmerge --force-merge-all -o ${project}.family.vcf.gz ${LCL5_vcf_gz} ${LCL6_vcf_gz} ${LCL7_vcf_gz} ${LCL8_vcf_gz}

zcat ${project}.family.vcf.gz | grep '##' > header
zcat ${project}.family.vcf.gz | grep -v '#' > body
echo -e "#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tLCL5\tLCL6\tLCL7\tLCL8" > name
cat header name body > ${project}.family.vcf

>>>

runtime {

Loading…
Annulla
Salva