Reference genomes with lots of things

find_refs.sh

#!/bin/bash

DATE=`date +%Y-%m-%d--%H-%M-%S`

export RefGenDir="/scratch/Reference_Genomes/Public"
export SoftwareDir="/scratch/gencore/software/bin/build_indices"

read -r -d '' TEMPLATE <<EOF
#HPC queue=adserial
#HPC commands_per_node=12
#HPC cpus_per_task=12
#HPC procs=12
EOF

############################
# Exported Functions 
############################
function gatk_index(){
	REF=$1
	DIR=$(dirname $1)
	BASE=$(basename $1)

	CWD=$(pwd)

	uncompress_index $REF

	DICT=`echo $BASE | sed 's/.fa//'`
	DICT=`echo $BASE | sed 's/.fna//'`
	DICT=`echo $BASE | sed 's/.fasta//'`

cat <<EOF
	java -jar /scratch/gencore/software/picard/2.0.1/picard.jar CreateSequenceDictionary \\
	REFERENCE=${BASE} \\
	OUTPUT=${DICT}.dict; \\
	cd $CWD 

EOF
}

function samtools_index(){
	REF=$1
	DIR=$(dirname $1)
	BASE=$(basename $1)

	CWD=$(pwd)

	uncompress_index $REF

cat <<EOF
	samtools faidx $BASE; \\
	cd $CWD 

EOF
}

function bowtie_index(){
	REF=$1
	DIR=$(dirname $1)
	BASE=$(basename $1)

	CWD=$(pwd)

	uncompress_index $REF

cat <<EOF
	bowtie2-build $BASE $BASE; \\
	cd $CWD

EOF
}

function bwa_index(){
	REF=$1
	DIR=$(dirname $1)
	BASE=$(basename $1)

	CWD=$(pwd)

	uncompress_index $REF

cat <<EOF
	/scratch/gencore/software/bin/bwa/0.7.12/bin/bwa index $BASE; \\
	cd $CWD

EOF
}

function uncompress_index(){
	REF=$1
	DIR=$(dirname $1)

cat <<EOF
cd $DIR && \\
EOF

if [ -f "${REF}.gz" ]; then
cat <<EOF
	gzip -d $REF.gz && \\
EOF
fi

}

export -f gatk_index
export -f samtools_index
export -f bwa_index
export -f bowtie_index
export -f uncompress_index

############################
# Main Functions
############################

function find_refs(){
	ext=$1
	index=$2
	process=$3

	find `pwd` -name *${ext}.gz -o -name *.${ext} -print0 | sed 's/.${ext}.gz/.${ext}/g'  | xargs -0 -I {}  bash -c "if [ ! -f "{}.${index}" ]; then ${process}_index {};  fi" >> "/scratch/gencore/software/bin/build_indices/build_${process}.sh" 

##Force Reindex

	#find `pwd` -name *${ext}.gz -o -name *.${ext} -print0 | sed 's/.${ext}.gz/.${ext}/g'  | xargs -0 -I {}  bash -c "${process}_index {}" >> "/scratch/gencore/software/bin/build_indices/build_${process}.sh" 
}

function process_jobs() {
	index=$1
	process=$2

	EXTS=('fa' 'fna' 'fasta')
	for i in "${EXTS[@]}"; do
		find_refs $i $index $process
	done
}

function submit_jobs(){
	process=$1
	module purge && module load NYUAD/2.0 apps anyenv
	pbsrunner.pl --infile "/scratch/gencore/software/bin/build_indices/build_${process}.sh" --outdir "/scratch/gencore/pbslogs/${DATE}_${process}_index" --jobname "${process}_index"
}


############################
# Main Create Indices
############################

############################
# Bowtie
############################
cd /scratch/Reference_Genomes/Public
process='bowtie'
index='1.bt2'

cat > ${SoftwareDir}/build_${process}.sh <<EOF
$TEMPLATE
#HPC module=NYUAD/2.0 apps tuxedo/2.0

EOF

process_jobs $index $process
submit_jobs $process 

############################
#BWA
############################
cd /scratch/Reference_Genomes/Public
process='bwa'
index='bwa'

cat > ${SoftwareDir}/build_${process}.sh <<EOF
$TEMPLATE
#HPC module=NYUAD/2.0 apps anyenv/1

EOF

process_jobs $index $process
submit_jobs $process 


############################
##GATK
############################
cd /scratch/Reference_Genomes/Public
process='gatk'
index='dict'

cat > ${SoftwareDir}/build_${process}.sh <<EOF
$TEMPLATE
#HPC module=NYUAD/2.0 gcc zlib openssl anyenv/1 jdk

EOF

process_jobs $index $process
submit_jobs $process 

############################
##Samtools
############################
process='samtools'
index='fai'

cd /scratch/Reference_Genomes/Public

cat > ${SoftwareDir}/build_${process}.sh <<EOF
$TEMPLATE
#HPC module=NYUAD/2.0 gcc zlib openssl anyenv/1 ncurses samtools

EOF

process_jobs $index $process
submit_jobs $process