#!/bin/bash

#SBATCH --array=125  # NEED TO CHANGE THIS!
#SBATCH --job-name=htstream # Job name
#SBATCH --nodes=1
#SBATCH --ntasks=4 # Number of cores
#SBATCH --time=6:0:0
#SBATCH --mem=16000 # Memory pool for all cores (see also --mem-per-cpu)
#SBATCH --partition=production # Partition to submit to
#SBATCH --output=arrayJob_%A_%a.out # File to which STDOUT will be written
#SBATCH --error=arrayJob_%A_%a.err # File to which STDERR will be written


start=`date +%s`

echo $HOSTNAME

echo "My SLURM_ARRAY_TASK_ID: " $SLURM_ARRAY_TASK_ID

THREADS=8

sample=`sed "${SLURM_ARRAY_TASK_ID}q;d" array_samples.txt`

echo $sample

inpath='/share/biocore/projects/Gershwin_L_UCD/Gershwin-Lebedev_Bos_taurus_Batch_TagSeq/00-RawData'
outpath='/share/biocore/projects/Gershwin_L_UCD/Gershwin-Lebedev_Bos_taurus_Batch_TagSeq/01-HTS_Preproc'

mkdir ${outpath}/${sample}

module load htstream/1.0.0

# export PATH=/share/biocore/software/bin:$PATH


hts_Stats -F -O \
        -L ${outpath}/${sample}/${sample}.htsStats.log \
        -U ${inpath}/${sample}/*.fastq.gz | \
hts_SeqScreener -F -S -O -A \
        -L ${outpath}/${sample}/${sample}.htsStats.log | \
hts_SeqScreener -F -S -O -A -r \
        -s /share/genomes/ensembl/Bos_taurus.ARS-UCD1.2/rRNA_NCBI/Bos_taurus_rRNA_2019.fasta \
        -L ${outpath}/${sample}/${sample}.htsStats.log | \
hts_AdapterTrimmer -F -S -O -A -m 50\
        -L ${outpath}/${sample}/${sample}.htsStats.log | \
hts_QWindowTrim -F -S -O -A -m 50 \
        -L ${outpath}/${sample}/${sample}.htsStats.log | \
hts_CutTrim -F -S -O -A -a 22 -m 40 \
        -L ${outpath}/${sample}/${sample}.htsStats.log | \
hts_Stats -F -S -A -L ${outpath}/${sample}/${sample}.htsStats.log \
        -g -p ${outpath}/${sample}/${sample}
        
exit


# To check:
for file in ./*/*.log ; do cat $file | grep totalFragmentsInput | head -1 | cut -f2 -d ":" | sed 's/ //' | sed 's/,//' ; done
for file in ./*/*.log ; do cat $file | grep totalFragmentsOutput | tail -1 | cut -f2 -d ":" | sed 's/ //' | sed 's/,//' ; done