NEBtest

#FASTQ_Me and CpG_Me with placenta samples that used NEB kit

cd /Unaligned/Project_L3_Nova062P

# Arguments
project=Pilot
#https://bioinformatics.ucdavis.edu/research-computing/documentation/archiving-slims-data/
SLIMSstring=604up50qc5
SLIMSdir=Unaligned/Project_L3_Nova062P
genome=hg38
programsPATH=/share/lasallelab/programs

echo "Creating a file of unique IDs based on first string from underscore delimiter"
ls -1 *fastq.gz | \
awk -F '_' '{print $1}' | \
sort -u > \
task_samples.txt

echo "Printing merge commands"
mergeLanesTest(){
	i=$1
	echo cat ${i}\_*_R1_001.fastq.gz \> ${i}\_1.fq.gz
	echo cat ${i}\_*_R2_001.fastq.gz \> ${i}\_2.fq.gz
}
export -f mergeLanesTest
cat task_samples.txt | parallel --jobs 6 --will-cite mergeLanesTest

echo "Merging ${lanes} lanes for ${R1} samples"
mergeLanes(){
	i=$1
	cat ${i}\_*_R1_001.fastq.gz > ${i}\_1.fq.gz
	cat ${i}\_*_R2_001.fastq.gz > ${i}\_2.fq.gz
}
export -f mergeLanes
cat task_samples.txt | parallel --jobs 6 --verbose --will-cite  mergeLanes

# Concise merging calls but doesn't make it easy for those who don't use GNU parallel
#parallel --jobs 6 "echo cat {}_*_R1_001.fastq.gz \> {}\_1.fq.gz" ::: `cat task_samples.txt`
#parallel --jobs 6 "echo cat {}_*_R2_001.fastq.gz \> {}\_2.fq.gz" ::: `cat task_samples.txt`
#parallel --jobs 6 --verbose "cat {}_"\*"_R1_001.fastq.gz > {}_1.fq.gz" ::: `cat task_samples.txt`
#parallel --jobs 6 --verbose "cat {}_"\*"_R2_001.fastq.gz > {}_2.fq.gz" ::: `cat task_samples.txt`

echo "Creating directories for CpG_Me"
mkdir ../../raw_sequences
mv task_samples.txt ../..
mv *.fq.gz ../../raw_sequences

echo "Checking read pairs"
cd ../../raw_sequences
pairedReads=$(($(ls | wc -l)/2))
if [ ${pairedReads} = ${R1} ]
then
        echo "${pairedReads} will be aligned using CpG_Me"
else
        echo "ERROR: Incorrect number of merged read pairs in raw_sequences"
        echo "There are ${pairedReads} but there should be ${R1}"
        exit 1
fi

echo "Submitting $genome alignment command to cluster for ${project}"
cd ..
call="sbatch \
--array=1-${pairedReads} \
${programsPATH}/CpG_Me/Paired-end/CpG_Me_PE_controller.sh \
${genome}"
	
echo ${call}
eval ${call}

echo "Done"
exit 0


cd /Unaligned/Project_L3_Nova062P
sbatch /CpG_Me/Paired-end/CpG_Me_PE_QC.sh

cd NEBtest/604up50qc5

# Arguments
project=Pilot
#https://bioinformatics.ucdavis.edu/research-computing/documentation/archiving-slims-data/
SLIMSstring=604up50qc5
SLIMSdir=Unaligned/Project_L3_Nova062P
genome=hg38
programsPATH=/share/lasallelab/programs

echo "Submitting $genome alignment command to cluster for ${project}"
sbatch --array=2 /share/lasallelab/programs/CpG_Me/Paired-end/CpG_Me_PE_controller.sh hg38
	
echo ${call}
eval ${call}

echo "Done"
exit 0