#!/bin/bash
#SBATCH --mail-user=adalena.nanni@ufl.edu
#SBATCH --mail-type=FAIL
#SBATCH	--job-name=bam2fq
#SBATCH	--output=/blue/mcintyre/share/transcript_distance/scripts/dros_pacbio/SLURM_LOGS/bam2fq_%A_%a.out
#SBATCH --time=01:00:00
#SBATCH --ntasks=1
#SBATCH --mem=3gb
#SBATCH --cpus-per-task=1
#SBATCH --array=1-8
#SBATCH --dependency=afterok:39750464

## convert bam after refining to fastq files
##     for aligning to reference genome with minimap

module load samtools/1.10 bedtools/2.29.2

### Set Directories
PROJ=/blue/mcintyre/share/transcript_distance
OUTD=$PROJ/dros_analysis/bam2fq
    mkdir -p ${OUTD}
IND=$PROJ/dros_analysis/isoseq3_library_processing

## Get info from design file

DESIGN_FILE=$PROJ/design_files/design_ethanol_srna_PacBio_path_noHeader_02avn.csv
DESIGN=$(sed -n "${SLURM_ARRAY_TASK_ID}p" $DESIGN_FILE)
IFS=',' read -ra ARRAY <<< "$DESIGN"

PACBIO_NUM=${ARRAY[0]}
SPECIES=${ARRAY[1]}
GENO=${ARRAY[2]}
SEX=${ARRAY[3]}
TRT=${ARRAY[4]}

SAMPLEID=${SPECIES}_${SEX}_${TRT}_${PACBIO_NUM}

date

echo "
*** SAM to FQ***

Input: ${IND}/${SAMPLEID}_ccs_trimmed.flnc.bam
"
bedtools bamtofastq \
    -i ${IND}/${SAMPLEID}_ccs_trimmed.flnc.bam \
    -fq ${OUTD}/${SAMPLEID}_ccs_trimmed.flnc.fq

date