| 1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889 |
- #!/bin/sh
- #PBS -N noUMI_varscan_pair
- #PBS -j oe
- #PBS -l ncpus=12
- #PBS -l nodes=1
- #PBS -l mem=20G
- #inputpath=$1
- #sample=$2
- source /home/liuxiangqiong/miniconda3/bin/activate base
- hg19=/cgdata/Database/GATK/b37/human_g1k_v37_decoy.fasta
- target=/cgdata/liuxiangqiong/work62pancancer/pipeline/v0/refdata/NanOnco_Plus_Panel_v2.0_Covered_b37_cg.parY2X.sort
- #min_af=0.001
- Vardict_Dir=/cgdata/liuxiangqiong/software/VarDict-1.5.7
- outputdir=${inputpath}/1SV_vardict_pair
- tumorbam=${bam_dir}/${tumor}_clean.bam
- normalbam=${bam_dir}/${normal}_clean.bam
- #pair
- for i in `seq 1 8`
- do
- ${Vardict_Dir}/vardict -G ${hg19} -f ${min_af} -N ${sample} -b "${tumorbam}|${normalbam}" -c 1 -S 2 -E 3 -g 4 ${target}-$i.bed |${Vardict_Dir}/testsomatic.R|${Vardict_Dir}/var2vcf_paired.pl -N "${tumor}|${normal}" -f ${min_af} > ${outputdir}/${sample}_pair.vd_$i.vcf&
- done
- wait
- #bgzip
- bcftools view ${outputdir}/${sample}_pair.vd_1.vcf -Oz -o ${outputdir}/${sample}_pair.vd_1.vcf.gz
- bcftools view ${outputdir}/${sample}_pair.vd_2.vcf -Oz -o ${outputdir}/${sample}_pair.vd_2.vcf.gz
- bcftools view ${outputdir}/${sample}_pair.vd_3.vcf -Oz -o ${outputdir}/${sample}_pair.vd_3.vcf.gz
- bcftools view ${outputdir}/${sample}_pair.vd_4.vcf -Oz -o ${outputdir}/${sample}_pair.vd_4.vcf.gz
- bcftools view ${outputdir}/${sample}_pair.vd_5.vcf -Oz -o ${outputdir}/${sample}_pair.vd_5.vcf.gz
- bcftools view ${outputdir}/${sample}_pair.vd_6.vcf -Oz -o ${outputdir}/${sample}_pair.vd_6.vcf.gz
- bcftools view ${outputdir}/${sample}_pair.vd_7.vcf -Oz -o ${outputdir}/${sample}_pair.vd_7.vcf.gz
- bcftools view ${outputdir}/${sample}_pair.vd_8.vcf -Oz -o ${outputdir}/${sample}_pair.vd_8.vcf.gz
- #index
- bcftools index ${outputdir}/${sample}_pair.vd_1.vcf.gz
- bcftools index ${outputdir}/${sample}_pair.vd_2.vcf.gz
- bcftools index ${outputdir}/${sample}_pair.vd_3.vcf.gz
- bcftools index ${outputdir}/${sample}_pair.vd_4.vcf.gz
- bcftools index ${outputdir}/${sample}_pair.vd_5.vcf.gz
- bcftools index ${outputdir}/${sample}_pair.vd_6.vcf.gz
- bcftools index ${outputdir}/${sample}_pair.vd_7.vcf.gz
- bcftools index ${outputdir}/${sample}_pair.vd_8.vcf.gz
- #
- bcftools concat -a ${outputdir}/${sample}_pair.vd_1.vcf.gz ${outputdir}/${sample}_pair.vd_2.vcf.gz ${outputdir}/${sample}_pair.vd_3.vcf.gz ${outputdir}/${sample}_pair.vd_4.vcf.gz ${outputdir}/${sample}_pair.vd_5.vcf.gz ${outputdir}/${sample}_pair.vd_6.vcf.gz ${outputdir}/${sample}_pair.vd_7.vcf.gz ${outputdir}/${sample}_pair.vd_8.vcf.gz | bcftools sort - -O z -o ${outputdir}/${sample}_pair.vd.vcf
- #标准化对齐
- vt decompose ${outputdir}/${sample}_pair.vd.vcf |vt normalize - -r ${hg19} -o ${outputdir}/${sample}_pair.vd.normalized.vcf
- #less ${outputdir}/${sample}_pair.vd.normalized.vcf | grep -vE "^#" > ${outputdir}/${sample}_pair.vd.normalized.nohead.vcf
- #3.3 将邻近的位点进行合并#3.3 将邻近的位点进行合并
- #python3 /cgdata/liuxiangqiong/work62pancancer/Client/v0/script/complex_variant_new.py -i ${outputdir}/${sample}_pair.vd.normalized.nohead.vcf -b ${bam_dir}/${tumor}_clean.bam -f ${hg19} -mode 2 -n 10 -p 0.5 -o ${outputdir}/${sample}_pair.vd.normalized.nohead.combind.vcf
- #grep "^#" ${outputdir}/${sample}_pair.vd.normalized.vcf >${outputdir}/${sample}_pair.vd.normalized.head.combind.vcf
- #less ${outputdir}/${sample}_pair.vd.normalized.nohead.combind.vcf >>${outputdir}/${sample}_pair.vd.normalized.head.combind.vcf
- conda deactivate
- #5.3对经过标准化的进行vep注释
- source /home/liuxiangqiong/miniconda3/bin/activate ensemble-vep
- PERL5LIB=/home/liuxiangqiong/miniconda3/envs/ensemble-vep/share/ensembl-vep-105.0-1/Bio/EnsEMBL/Variation/Utils/DB/HTS/Faidx
- export PERL5LIB=$PERL5LIB
- inputdir=${outputdir}/${sample}_pair.vd.normalized.vcf
- outdir1=${outputdir}/${sample}_pair.vd.normalized.vepanno.vcf
- vep --cache --fork 16 \
- -i ${inputdir} \
- -o ${outdir1} \
- --assembly GRCh37 \
- --cache_version 104 \
- --dir_cache /home/liuxiangqiong/miniconda3/share/ensembl-vep-88.9-0/hg19 \
- --refseq \
- --canonical \
- --offline \
- --hgvs \
- --fasta /cgdata/Database/GATK/b37/human_g1k_v37_decoy.fasta \
- --hgvsg \
- --max_af \
- --check_existing \
- --numbers \
- --symbol \
- --canonical \
- --offline \
- -vcf \
- --use_given_ref \
- --af_gnomad \
- --check_existing
- #5.4对vep的再进行annovar注释
- outdir2=${outputdir}/${sample}_pair.vd.normalized.vepanno.annovar
- perl /cgdata/CGTools/soft/src/annovar_20200608/table_annovar.pl ${outdir1} /cgdata/soft/src/annovar.archive/DB/ -buildver hg19 -out ${outdir2} -remove -protocol refGene,popfreq_max_20150413,gnomad_genome,avsnp150,snp138NonFlagged,dbnsfp33a,cosmic94,clinvar_20220829,intervar_20170202,icgc28 -operation g,f,f,f,f,f,f,f,f,f -nastring . -vcfinput --thread 40
- #remove
- rm ${outputdir}/${sample}_pair.vd_?.vcf*
|
