#!/bin/bash

#SBATCH --nodes=1
#SBATCH --ntasks-per-node=16
#SBATCH --mem=128GB
#SBATCH --time=12:00:00


# Personal stuff; Change this to your personal settings 
#SBATCH --job-name=Kraken2
#SBATCH --output=slurm-Kraken-%j.log

export KRAKEN2_DEFAULT_DB="/scratch/public/genomes/kraken2db"
module load Kraken2/2.1.2-gompi-2022a

cpu=$SLURM_JOB_CPUS_PER_NODE

##############################################   Functions   ######################################################################

function my_kraken {
   name=$1
   filename=$2
   kraken2 --threads $cpu --quick --report $resultsdir/$name.report $datadir/$filename > $resultsdir/$name.kraken.table
   rm -Rf $resultsdir/$name.kraken.table 
}

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

##############################################   Pipeline   ######################################################################

datadir=/scratch/hb-molgen/NGSdata/20240926_019
resultsdir=/scratch/p127804/PROJECTS/Douwe/Bifidobacterium_202409_019/results_kraken2

my_mkdir $resultsdir

# NOTE to get all filenames use :  ls -aR >TOC.txt

# my_kraken parameters
	# 1) name of your sample e.g. kraken_sample1
	# 2) fasta files including the path
	# e.g., my_kraken kraken_sample1 sample.fasta

# report headers; Percentage  Reads  Taxon            Rank        TaxID

# Percentage: The percentage of reads classified at that taxonomic level.
# Reads: The number of reads classified at that taxonomic level.
# Taxon: The taxonomic identifier or name of the organism at that level.
# Rank: The taxonomic rank of the classification (e.g., species, genus, family).
# TaxID: The unique identifier assigned to the taxon in the Kraken database.





	
# my_kraken Bd1 Bd1_6321_S84_R1_001.fastq.gz
# MM0213; 
my_kraken MM0213_DS019_01 DS019_01_S32_L001_R1_001.fastq.gz
my_kraken MM0213_DS019_02 DS019_02_S33_L001_R1_001.fastq.gz
my_kraken MM0213_DS019_03 DS019_03_S34_L001_R1_001.fastq.gz
my_kraken MM0213_DS019_04 DS019_04_S35_L001_R1_001.fastq.gz
my_kraken MM0213_DS019_05 DS019_05_S36_L001_R1_001.fastq.gz
my_kraken MM0213_DS019_06 DS019_06_S37_L001_R1_001.fastq.gz
my_kraken MM0213_DS019_07 DS019_07_S38_L001_R1_001.fastq.gz
my_kraken MM0213_DS019_08 DS019_08_S39_L001_R1_001.fastq.gz
my_kraken MM0213_DS019_09 DS019_09_S40_L001_R1_001.fastq.gz
my_kraken MM0213_DS019_10 DS019_10_S41_L001_R1_001.fastq.gz
my_kraken MM0213_DS019_11 DS019_11_S42_L001_R1_001.fastq.gz
my_kraken MM0213_DS019_12 DS019_12_S43_L001_R1_001.fastq.gz

# MM0160
my_kraken MM0160_DS019_13 DS019_13_S44_L001_R1_001.fastq.gz
my_kraken MM0160_DS019_14 DS019_14_S45_L001_R1_001.fastq.gz
my_kraken MM0160_DS019_15 DS019_15_S46_L001_R1_001.fastq.gz
my_kraken MM0160_DS019_16 DS019_16_S47_L001_R1_001.fastq.gz
my_kraken MM0160_DS019_17 DS019_17_S48_L001_R1_001.fastq.gz
my_kraken MM0160_DS019_18 DS019_18_S49_L001_R1_001.fastq.gz
my_kraken MM0160_DS019_19 DS019_19_S50_L001_R1_001.fastq.gz
my_kraken MM0160_DS019_20 DS019_20_S51_L001_R1_001.fastq.gz
my_kraken MM0160_DS019_21 DS019_21_S52_L001_R1_001.fastq.gz
my_kraken MM0160_DS019_22 DS019_22_S53_L001_R1_001.fastq.gz
my_kraken MM0160_DS019_23 DS019_23_S54_L001_R1_001.fastq.gz
my_kraken MM0160_DS019_24 DS019_24_S55_L001_R1_001.fastq.gz
