|
|
@@ -22,7 +22,7 @@ task VEP { |
|
|
|
|
|
|
|
source /etc/profile |
|
|
|
|
|
|
|
awk -F'\t' '{if(($1~"^#")||($1!~"^#" && $7=="PASS")){print $0}}' ${vcf} > ${sample_id}.vcf |
|
|
|
#awk -F'\t' '{if(($1~"^#")||($1!~"^#" && $7=="PASS")){print $0}}' ${vcf} > ${sample_id}.vcf |
|
|
|
|
|
|
|
# Judge the SAMPLE info of vcf file |
|
|
|
tumor_id=`awk -F'\t' '{if($1~"^#CHROM"){print $10}}' ${vcf}` |
|
|
@@ -36,18 +36,20 @@ task VEP { |
|
|
|
SAMPLE_vcf2vcf="--vcf-tumor-id $tumor_id" |
|
|
|
fi |
|
|
|
|
|
|
|
|
|
|
|
# Set the buffer_size based on the data size |
|
|
|
nrow=`awk -F'\t' '{if($1~"^chr"){print $0}}' ${sample_id}.vcf | wc -l` |
|
|
|
nrow=`awk -F'\t' '{if($1~"^chr"){print $0}}' ${vcf} | wc -l` |
|
|
|
if [ $nrow -lt 5000 ]; then |
|
|
|
buffer_size="--buffer_size 5000" |
|
|
|
else |
|
|
|
buffer_size="--buffer_size 1000" |
|
|
|
fi |
|
|
|
|
|
|
|
|
|
|
|
# vcf2vcf: transfer into a standardized format |
|
|
|
echo "Transfer the VCF file into a standardized format..." |
|
|
|
perl ${vcf2maf_path}/vcf2vcf.pl \ |
|
|
|
--input-vcf ${sample_id}.vcf --output-vcf ${basename}.norm.vcf \ |
|
|
|
--input-vcf ${vcf} --output-vcf ${basename}.norm.vcf \ |
|
|
|
$SAMPLE_vcf2vcf \ |
|
|
|
--ref-fasta ${ref_dir}/${fasta} |
|
|
|
|
|
|
@@ -62,6 +64,7 @@ task VEP { |
|
|
|
--cache --dir_cache ${cache} \ |
|
|
|
--fasta ${ref_dir}/${fasta} \ |
|
|
|
$buffer_size \ |
|
|
|
--fork $nt \ |
|
|
|
--input_file ${basename}.norm.vcf --output_file ${basename}.vep.vcf |
|
|
|
|
|
|
|
|