From ec1a962cb9fe101268cd03d97afade8c0f2942be Mon Sep 17 00:00:00 2001
From: Brandi Cantarel <brandi.cantarel@utsouthwestern.edu>
Date: Fri, 31 Jul 2020 11:13:23 -0500
Subject: [PATCH] update snpeff ref; remove unnecessary ref data checks
---
variants/germline_vc.sh | 25 ++++++++-----------------
variants/somatic_vc.sh | 23 ++++++-----------------
variants/svcalling.sh | 3 ++-
3 files changed, 16 insertions(+), 35 deletions(-)
diff --git a/variants/germline_vc.sh b/variants/germline_vc.sh
index cd1a1c7..0dc3659 100755
--- a/variants/germline_vc.sh
+++ b/variants/germline_vc.sh
@@ -1,4 +1,4 @@
-#!/bin/bash
+nucleu#!/bin/bash
#germline_vc.sh
usage() {
@@ -36,21 +36,7 @@ if [[ -z $NPROC ]]
then
NPROC=`nproc`
fi
-if [[ -s "${index_path}/dbSnp.vcf.gz" ]]
-then
- dbsnp="${index_path}/dbSnp.vcf.gz"
- gatk4_dbsnp="${index_path}/dbSnp.gatk4.vcf.gz"
-else
- echo "Missing dbSNP File: ${index_path}/dbSnp.vcf.gz"
- usage
-fi
-if [[ -s "${index_path}/GoldIndels.vcf.gz" ]]
-then
- knownindel="${index_path}/GoldIndels.vcf.gz"
-else
- echo "Missing InDel File: ${index_path}/GoldIndels.vcf.gz"
- usage
-fi
+
if [[ -s "${index_path}/genome.fa" ]]
then
reffa="${index_path}/genome.fa"
@@ -112,6 +98,12 @@ then
bcftools norm -c s -f ${reffa} -w 10 -O z -o ${pair_id}.platypus.vcf.gz platypus.vcf.gz
elif [[ $algo == 'gatk' ]]
then
+ gatk4_dbsnp="${index_path}/dbSnp.gatk4.vcf.gz"
+ if [[ ! -f "${index_path}/dbSnp.gatk4.vcf.gz" ]]
+ then
+ echo "Missing dbSNP File: ${index_path}/dbSnp.vcf.gz"
+ usage
+ fi
user=$USER
module load gatk/4.1.4.0
gvcflist=''
@@ -129,7 +121,6 @@ then
tabix ${pair_id}.gatk.vcf.gz
elif [ $algo == 'mutect' ]
then
- gatk4_dbsnp=${index_path}/clinseq_prj/dbSnp.gatk4.vcf.gz
module load gatk/4.1.4.0 parallel/20150122
threads=`expr $NPROC / 2`
bamlist=''
diff --git a/variants/somatic_vc.sh b/variants/somatic_vc.sh
index 94d4ece..d3701aa 100755
--- a/variants/somatic_vc.sh
+++ b/variants/somatic_vc.sh
@@ -56,7 +56,6 @@ if [[ -f $pon ]]
then
ponopt="--pon $pon"
fi
-
if [[ -a "${index_path}/genome.fa" ]]
then
reffa="${index_path}/genome.fa"
@@ -65,21 +64,7 @@ else
echo "Missing Fasta File: ${index_path}/genome.fa"
usage
fi
-if [[ -a "${index_path}/dbSnp.vcf.gz" ]]
-then
- dbsnp="${index_path}/dbSnp.vcf.gz"
-else
- echo "Missing dbSNP File: ${index_path}/dbSnp.vcf.gz"
- usage
-fi
-if [[ -a "${index_path}/cosmic.vcf.gz" ]]
-then
- cosmic=${index_path}/cosmic.vcf.gz
-else
- echo "Missing InDel File: ${index_path}/cosmic.vcf.gz"
- usage
-fi
baseDir="`dirname \"$0\"`"
interval=`cat ${reffa}.fai |cut -f 1 |grep -v decoy |grep -v 'HLA' |grep -v alt |grep -v 'chrUn' |grep -v 'random' | perl -pe 's/\n/ -L /g' |perl -pe 's/-L $//'`
@@ -96,7 +81,6 @@ then
nid=`samtools view -H ${normal} |grep '^@RG' |perl -pi -e 's/\t/\n/g' |grep ID |cut -f 2 -d ':'`
fi
-
if [ $algo == 'strelka2' ]
then
module load strelka/2.9.10 manta/1.3.1
@@ -127,6 +111,12 @@ then
elif [ $algo == 'virmid' ]
then
module load virmid/1.2
+ cosmic=${index_path}/cosmic.vcf.gz
+ if [[ ! -f "${index_path}/cosmic.vcf.gz" ]]
+ then
+ echo "Missing InDel File: ${index_path}/cosmic.vcf.gz"
+ usage
+ fi
virmid -R ${reffa} -D ${tumor} -N ${normal} -s ${cosmic} -t $NPROC -M 2000 -c1 10 -c2 10
perl $baseDir/addgt_virmid.pl ${tumor}.virmid.som.passed.vcf
perl $baseDir/addgt_virmid.pl ${tumor}.virmid.loh.passed.vcf
@@ -135,7 +125,6 @@ then
vcf-concat *gt.vcf | vcf-sort | vcf-annotate -n --fill-type -n | java -jar $SNPEFF_HOME/SnpSift.jar filter '((NDP >= 10) & (DDP >= 10))' | perl -pe "s/TUMOR/${tid}/g" | perl -pe "s/NORMAL/${nid}/g" | bgzip > ${pair_id}.virmid.vcf.gz
elif [ $algo == 'mutect' ]
then
- gatk4_dbsnp=${index_path}/clinseq_prj/dbSnp.gatk4.vcf.gz
module load gatk/4.1.4.0 parallel/20150122
threads=`expr $NPROC / 2`
gatk --java-options "-Xmx20g" Mutect2 $ponopt --independent-mates -RF AllowAllReadsReadFilter -R ${reffa} -I ${tumor} -tumor ${tid} -I ${normal} -normal ${nid} --output ${tid}.mutect.vcf -L $interval
diff --git a/variants/svcalling.sh b/variants/svcalling.sh
index a14ff60..e030948 100755
--- a/variants/svcalling.sh
+++ b/variants/svcalling.sh
@@ -11,7 +11,7 @@ usage() {
exit 1
}
OPTIND=1 # Reset OPTIND
-while getopts :r:p:b:t:x:c:y:n:l:a:hf opt
+while getopts :r:p:b:t:x:c:g:y:n:l:a:hf opt
do
case $opt in
r) index_path=$OPTARG;;
@@ -22,6 +22,7 @@ do
x) tid=$OPTARG;;
y) nid=$OPTARG;;
f) filter=1;;
+ g) snpeffgeno=$OPTARG;;
b) sbam=$OPTARG;;
c) tbed=$OPTARG;;
l) itdbed=$OPTARG;;
--
GitLab