#!/bin/bash
#SBATCH --dependency=afterany:4873:4874:4875:4876:4877
set -e
set -o pipefail
umask 0002

#### SET THE STAGE
SCRATCH_DIR=/scratch/DNABamStats_2024-11-01--10-45-40_NA12892_chr10_temp$$
GSTORE_DIR=/srv/GT/analysis/course_sushi/public/gstore/projects
INPUT_DATASET=/srv/GT/analysis/course_sushi/public/gstore/projects/p1007/DNABamStats_2024-11-01--10-45-40/input_dataset.tsv
LAST_JOB=TRUE
echo "Job runs on `hostname`"
echo "at $SCRATCH_DIR"
mkdir $SCRATCH_DIR || exit 1
cd $SCRATCH_DIR || exit 1
source /usr/local/ngseq/etc/lmod_profile
module add Dev/R/4.4.0 Tools/Picard/3.2.0 QC/Qualimap/2.3.0 QC/SAMStat/1.5.1 Tools/samtools/1.20

#### NOW THE ACTUAL JOBS STARTS
#echo Display=$DISPLAY
set -x
REF=/srv/GT/reference/Homo_sapiens/Ensembl/GRCh37.p13/Annotation/Version-2015-03-05/../../Sequence/WholeGenomeFasta/genome.fa
BAM_FILE=/srv/GT/analysis/course_sushi/public/gstore/projects/p1007/Bowtie2_myPre_WES_all_2024-11-01--10-41-52/NA12892_chr10.bam
FN=$(basename $BAM_FILE)
###samstat
cp -a $BAM_FILE .
samstat $FN
mv {$FN,NA12892_chr10}.samstat.html
###qualimap
unset DISPLAY ; qualimap bamqc -bam $FN -c -nt 8 --java-mem-size=30G -outdir NA12892_chr10 > qualimap.out
###picard 
/usr/local/ngseq/packages/Dev/jdk/21/bin/java -Xmx30g -jar $Picard_jar CollectGcBiasMetrics OUTPUT=NA12892_chr10.gc.dat SUMMARY_OUTPUT=NA12892_chr10.gc.sum.dat INPUT=$FN CHART_OUTPUT=NA12892_chr10.picard.pdf ASSUME_SORTED=true REFERENCE_SEQUENCE=$REF VALIDATION_STRINGENCY=LENIENT > picard.out 2> picard.err


#### JOB IS DONE WE PUT THINGS IN PLACE AND CLEAN AUP
rsync -r NA12892_chr10.samstat.html /srv/GT/analysis/course_sushi/public/gstore/projects/p1007/DNABamStats_2024-11-01--10-45-40/
rsync -r NA12892_chr10 /srv/GT/analysis/course_sushi/public/gstore/projects/p1007/DNABamStats_2024-11-01--10-45-40/
rsync -r NA12892_chr10.picard.pdf /srv/GT/analysis/course_sushi/public/gstore/projects/p1007/DNABamStats_2024-11-01--10-45-40/
cd /scratch
rm -rf /scratch/DNABamStats_2024-11-01--10-45-40_NA12892_chr10_temp$$ || exit 1


echo __SCRIPT END__