runBulkCHIP_fastq.pbs

#!/bin/bash
#PBS -q condo
#PBS -N chiseq
#PBS -l nodes=1:ppn=16
#PBS -l walltime=8:00:00
#PBS -V
#PBS -m abe
#PBS -A epigen-group

source activate aquas_chipseq

WORKDIR="/oasis/tscc/scratch/$(whoami)/outputs/"
FASTQDIR="/projects/ps-epigen/seqdata/"

# chipseq or not
[[ "$chipseq" = true ]] &&  c="-type chip-seq"

# select libs col1:libid,col2:genome,col3:PE/SE
ncol=$(awk 'END{print NF}' $samples)
[[ $ncol -ne 4 ]] && { echo "sample file dosen't have 4 columns "; exit 0; }

samplenames=(`cat $samples`)
INPREFIX=${samplenames[${PBS_ARRAYID}*4]} #index start from 0
GENOME=${samplenames[${PBS_ARRAYID}*4+1]}
READ=${samplenames[${PBS_ARRAYID}*4+2]}
IS_INPUT=${samplenames[${PBS_ARRAYID}*4+3]}

# split name into subnames
subnames=($(echo $INPREFIX|awk -v FS='_' '{for(i=1;i<=NF;i++) print $i}'))
n_subnames=${#subnames[@]} # number of subnames split at _
snap='true' #default turn on snap-chip 

# parameters
type="histone"
true_rep="true"
no_pseudo_rep="true"

# output dir 
OUTDIR="${WORKDIR}${INPREFIX}"
mkdir -p $OUTDIR

if [ $READ = 'PE' ] 
then
    echo "Using PE pipepline"

    # check fastqs
    fastq1="${FASTQDIR}/${INPREFIX}_R1.fastq.gz"
    fastq2="${FASTQDIR}/${INPREFIX}_R2.fastq.gz"
    echo $fastq1

    # no fastqs and not merged libs 
    [[ ! -f $fastq1 ]] && [[ $n_subnames -lt 4 ]] && { echo "$fastq1 fastq1 not found"; exit 0; }
    [[ ! -f $fastq2 ]] && [[ $n_subnames -lt 4 ]] && { echo "$fastq1 fastq2 not found"; exit 0; }
    
    ##  merge submembers if submembers presents
    if [ ! -f $fastq1 ] || [ ! -f $fastq2 ] # break down to check components
    then       
       base_id=${subnames[0]}_${subnames[1]}
       tmp=(${subnames[@]/#/${base_id}_})
       sub_libs=($(echo ${tmp[@]:2}| sed 's/_1 / /g'))
       tmp=($(echo ${sub_libs[@]./#/${FASTQDIR}}))
       tmp_r1=${tmp[@]/%/_R1.fastq.gz}
       tmp_r2=${tmp[@]/%/_R2.fastq.gz}              
       [[ ! -f $fastq1 ]] && { echo "cat $tmp_r1 > $fastq1 & sleep 1" | bash;}
       [[ ! -f $fastq2 ]] && { echo "cat $tmp_r2 > $fastq2 & sleep 1" | bash;}
    fi
    wait 

    # runPipeline
    bds /projects/ps-epigen/software/atac_dnase_pipelines/atac.bds -species $GENOME -nth $PBS_NP \
        -fastq1_1 "$fastq1" \
        -fastq1_2 "$fastq2" \
        -true_rep -no_idr  -no_par -no_xcor \
        -out_dir $OUTDIR $c
    wait

    ## addtional check
    [[ -z $(find ${OUTDIR}/qc/rep1 -name "*_qc.txt") ]]  && { echo  "pipeline failed"; exit 0; }

    # runFastQC & fastq_screen
    runFastQC_screen.sh  $INPREFIX &

    ## snap-chip 
    #export PATH=$PATH:/projects/ps-epigen/software/SNAP-CHIP_epicypher
    #bash epicypher.sh -i $fastq1 -m 2 -k true -o /projects/ps-epigen/outputs/libQCs/${INPREFIX}/ 1 | tee /projects/ps-epigen//outputs/libQCs/${INPREFIX}/${INPREFIX}_R1.fastq.gz.snap.cnt.tab &
    #bash epicypher.sh -i $fastq2 -m 2 -k true -o /projects/ps-epigen/outputs/libQCs/${INPREFIX}/ 1 | tee /projects/ps-epigen//outputs/libQCs/${INPREFIX}/${INPREFIX}_R2.fastq.gz.snap.cnt.tab

else
    echo "Using SE pipepline"

    # fastq check 
    fq="${FASTQDIR}/${INPREFIX}.fastq.gz"
    [[ ! -f $fq ]] && [[ $n_subnames -lt 4 ]] &&  { echo "$fq fastq not found"; exit 0; }

    ##  merge submembers if submembers presents
    if [ ! -f $fq ] 
    then       
       base_id=${subnames[0]}_${subnames[1]}
       tmp=(${subnames[@]/#/${base_id}_})
       sub_libs=($(echo ${tmp[@]:2}| sed 's/_1 / /g'))
       tmp=($(echo ${sub_libs[@]./#/${FASTQDIR}}))
       tmp_r1=(${tmp[@]/%/_R1.fastq.gz})
       i=0;for fl in ${tmp_r1[@]};do [[ ! -f $fl ]] && tmp_r1[$i]=${fl/_R1./.};i=$[i+1];echo $i; done 
       echo "cat ${tmp_r1[@]} > $fq & sleep 1" | bash
    fi
    wait 
    
    # run pipeline
    bds /projects/ps-epigen/software/atac_dnase_pipelines/atac.bds -species $GENOME -nth $PBS_NP \
        -se -fastq1 $fq \
        -true_rep -no_idr  -no_par \
        -out_dir $OUTDIR $c
    wait

    ## addtional check
    [[ -z $(find ${OUTDIR}/qc/rep1 -name "*_qc.txt") ]]  && { echo  "pipeline failed"; exit 0; }
    
    # fastqc and screen
    runFastQC_screen_se.sh $INPREFIX &

    ## snap-chip 
    #export PATH=$PATH:/projects/ps-epigen/software/SNAP-CHIP_epicypher
    #bash epicypher.sh -i $fq -m 2 -k true -o /projects/ps-epigen/outputs/libQCs/${INPREFIX}/ 1 | tee /projects/ps-epigen//outputs/libQCs/${INPREFIX}/${INPREFIX}.fastq.gz.snap.cnt.tab &
    
fi

wait

# result transfer 
results_transfer.sh $INPREFIX $(whoami)

# add .finished.txt tag
touch /projects/ps-epigen/outputs/libQCs/${INPREFIX}/.finished.txt