Pipeline for NBN ChIP-seq

nbn-chipseq-workflow.sh

# alias the raw data files
ln -s C0PYJACXX_s7_0_GSL48index_10_SL16067.fastq.gz 0h-IP-nbn.fq.gz
ln -s C0PYJACXX_s7_0_GSL48index_11_SL16068.fastq.gz 3Gy-IP-nbn.fq.gz
ln -s C0PYJACXX_s7_0_GSL48index_12_SL16069.fastq.gz PpoI-IP-nbn.fq.gz
ln -s C0PYJACXX_s7_0_GSL48index_7_SL16064.fastq.gz 0h-Input.fq.gz
ln -s C0PYJACXX_s7_0_GSL48index_8_SL16065.fastq.gz 3Gy-Input.fq.gz
ln -s C0PYJACXX_s7_0_GSL48index_9_SL16066.fastq.gz PpoI-Input.fq.gz

# build a list of sample stubs
SAMPLES="0h-Input 0h-IP-nbn 3Gy-Input 3Gy-IP-nbn PpoI-Input PpoI-IP-nbn"

# align the raw FASTQ files
STEPNAME=bwa-aln
GENOME=~/shared/genomes/hg19/bwa/gatk/hg19_gatk.fa
WHERE=/home/arq5x/cphg-home/projects/concannon-nibrin-chipseq/orig-fastqs-from-ha
export QSUB="qsub -q cphg -W group_list=CPHG -V -l select=1:mem=4000m:ncpus=8 -N $STEPNAME -m bea -M [email protected]"
for sample in $SAMPLES
do
   echo "cd $WHERE; bwa aln -t 6 $GENOME $sample.fq.gz > $sample.sai" | $QSUB
done 


# BWA SAMSE
STEPNAME=bwa-samse
GENOME=~/shared/genomes/hg19/bwa/gatk/hg19_gatk.fa
WHERE=/home/arq5x/cphg-home/projects/concannon-nibrin-chipseq
export QSUB="qsub -q cphg -W group_list=CPHG -V -l select=1:mem=4000m:ncpus=8 -N $STEPNAME -m bea -M [email protected]"
for sample in $SAMPLES
do
   echo "cd $WHERE; bwa samse -n 100 -r '@RG\tID:$sample\tSM:$sample' $GENOME \
                                         orig-fastqs-from-ha/$sample.sai \
                                         orig-fastqs-from-ha/$sample.fq.gz \
                    | samtools view -Sb - > bam/$sample.bam" \
                    | $QSUB
done 



# samtools sort
STEPNAME=smtls-srt
WHERE=/home/arq5x/cphg-home/projects/concannon-nibrin-chipseq
export QSUB="qsub -q cphg -W group_list=CPHG -V -l select=1:mem=4000m:ncpus=2 -N $STEPNAME -m bea -M [email protected]"
for sample in $SAMPLES
do
   echo "cd $WHERE; samtools sort bam/$sample.bam bam/$sample.possort" | $QSUB
done 


# samtools index
STEPNAME=smtls-idx
WHERE=/home/arq5x/cphg-home/projects/concannon-nibrin-chipseq
export QSUB="qsub -q cphg -W group_list=CPHG -V -l select=1:mem=4000m:ncpus=2 -N $STEPNAME -m bea -M [email protected]"
for sample in $SAMPLES
do
   echo "cd $WHERE; samtools index bam/$sample.possort.bam" | $QSUB
done

# make a MAPQ >=30 BEDGRAPH for each sample
STEPNAME=bedgraph
WHERE=/home/arq5x/cphg-home/projects/concannon-nibrin-chipseq
export QSUB="qsub -q cphg -W group_list=CPHG -V -l select=1:mem=4000m:ncpus=2 -N $STEPNAME -m bea -M [email protected]"
for sample in $SAMPLES
do
   echo "cd $WHERE; samtools view -uq 30 bam/$sample.possort.bam | \
                   bedtools genomecov -ibam - -bg > bam/$sample.possort.bam.q30.bedg" | $QSUB
done