#!/bin/bash

# https://www.bioinformatics.babraham.ac.uk/projects/fastqc/

# =============================================================================================================================================================
# ==                                                          SLURM parameters                                                                               ==
# =============================================================================================================================================================

#SBATCH --nodes=1			
#SBATCH --ntasks-per-node=4
#SBATCH --mem=8GB
#SBATCH --time=24:00:00
#SBATCH --job-name=FastQC
#SBATCH --output=slurm-Anne-%j.log


# =============================================================================================================================================================
# ==                                                          Load modules                                                                               ==
# =============================================================================================================================================================

module load FastQC/0.11.9-Java-11
module load Trimmomatic/0.39-Java-11
module load Trim_Galore/0.6.6-GCCcore-9.3.0-Python-3.8.2
cpu=4

# =============================================================================================================================================================
# ==                                                          Functions                                                                               ==
# =============================================================================================================================================================

function my_fastq_readcount {
	# NOTE: not used in this script
	name=$1
	fastq=$2
	echo $name $(zgrep -c "^+$" $datadir/$fastq) >> $resultsdir/00.fastq.counts
}

function my_fastqc {
	name=$1
	fastq=$2
	printf "\n========== FastQC: "$name" ======================================\n"	
	my_mkdir $resultsdir/$name
	fastqc -t $cpu -o $resultsdir/$name -f fastq $datadir/$fastq
}

function trimmomatic {
	name=$1
	fastq=$2
	printf "\n========== Trimmomatic: "$name" ======================================\n"
	#java -jar $EBROOTTRIMMOMATIC/trimmomatic-0.39.jar SE -phred33 -threads $cpu $datadir/$fastq $datadir/$name.$fastq ILLUMINACLIP:/data/pg-molgen/software/trimmomatic/TruSeq3-PE.fa:2:30:10 LEADING:3 TRAILING:3 SLIDINGWINDOW:4:15 MINLEN:36
	java -jar $EBROOTTRIMMOMATIC/trimmomatic-0.39.jar SE -phred33 -threads $cpu $datadir/$fastq $datadir/$name.36.$fastq CROP:36
}




function my_mkdir {
	if [ ! -d $1 ]; then
		mkdir $1 ;
	fi
}


# =============================================================================================================================================================
# ==                                                          Main script                                                                               ==
# =============================================================================================================================================================



datadir=/scratch/p127804/PROJECTS/PrimBio/20240113/fastq
resultsdir=/scratch/p127804/PROJECTS/PrimBio/20240113/fastqc

my_mkdir $resultsdir

# below you see: 1) function, 2) name of the sample and 3) the fastq filename
	
my_fastqc Bd1 Bd1_6321_S84_R1_001.fastq.gz
my_fastqc Cm10 Cm10_6291_S54_R1_001.fastq.gz