spacesavers2_e2e

#!/usr/bin/env bash
####################################################################################
# spacesavers2 end-to-end wrapper script
####################################################################################
set -e -o pipefail

SCRIPT_DIR=$( cd -- "$( dirname -- "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )

# ncpus=`nproc`

ARGPARSE_DESCRIPTION="End-to-end run of spacesavers2"
source ${SCRIPT_DIR}/resources/argparse.bash || exit 1
argparse "$@" <<EOF || exit 1
parser.add_argument('-f','--folder',required=True, help='Folder to run spacesavers_catalog on.')
parser.add_argument('-p','--threads',required=False, help='number of threads to use', default=4)
parser.add_argument('-d','--maxdepth',required=False, help='maxdepth for mimeo', default=4)
parser.add_argument('-l','--limit',required=False, help='limit for running spacesavers_grubbers', default=5)
parser.add_argument('-q','--quota',required=False, help='total size of the volume (default = 200 for /data/CCBR)', default=200)
parser.add_argument('-o','--outfolder',required=True, help='Folder where all spacesavers_e2e output files will be saved')
EOF

# assuming that python 3.11 is available with xxhash module
# . "/data/CCBR_Pipeliner/db/PipeDB/Conda/etc/profile.d/conda.sh" && conda activate py311

if [ "$SLURM_JOB_ID" == "" ]; then 
    tmpdir="$OUTFOLDER"
else
    tmpdir="/lscratch/${SLURM_JOB_ID}"
fi

# run spacesavers2
dt=$(date +%Y%m%d)
prefix=$(echo "${dt}.${FOLDER}"|sed "s/\//_/g")
outfile_catalog="${OUTFOLDER}/${prefix}.catalog"
outfile_catalog_err="${OUTFOLDER}/${prefix}.catalog.err"
outfile_catalog_log="${OUTFOLDER}/${prefix}.catalog.log"
outfile_mimeo_log="${OUTFOLDER}/${prefix}.mimeo.log"
outfile_mimeo_err="${OUTFOLDER}/${prefix}.mimeo.err"

outfile_blamematrix="${OUTFOLDER}/${prefix}.blamematrix.tsv"
outfile_blamematrix_log="${OUTFOLDER}/${prefix}.blamematrix.log"
outfile_blamematrix_err="${OUTFOLDER}/${prefix}.blamematrix.err"

if [ ! -d $OUTFOLDER ];then mkdir -p $OUTFOLDER;fi

# spacesavers2_catalog
if [ "$?" == "0" ];then
echo "Running spacesavers2_catalog" && \
spacesavers2_catalog \
    --folder $FOLDER \
    --threads $THREADS \
    --outfile ${outfile_catalog} \
    --bottomhash \
    > ${outfile_catalog_log} 2> ${outfile_catalog_err}
fi

sleep 60

# spacesavers2_mimeo
if [ "$?" == "0" ];then
echo "Running spacesavers2_mimeo" 
command -V ktImportText 2>/dev/null || module load kronatools || (>&2 echo "module kronatools could not be loaded")
spacesavers2_mimeo \
    --catalog ${outfile_catalog} \
    --outdir ${OUTFOLDER} \
    --quota $QUOTA \
    --duplicatesonly \
    --maxdepth $MAXDEPTH \
    --p $prefix \
    --kronaplot \
    > ${outfile_mimeo_log} 2> ${outfile_mimeo_err}
fi

sleep 60

# spacesavers2_grubbers
if [ "$?" == "0" ];then
echo "Running spacesavers2_grubbers" && \
for filegz in `ls ${OUTFOLDER}/${prefix}*files.gz`;do
    outfile=`echo $filegz|sed "s/mimeo.files.gz/grubbers.tsv/g"`
    logfile=`echo $filegz|sed "s/mimeo.files.gz/grubbers.log/g"`
    errfile=`echo $filegz|sed "s/mimeo.files.gz/grubbers.err/g"`
    spacesavers2_grubbers \
        --filesgz $filegz \
        --limit $LIMIT \
        --outfile $outfile \
        > $logfile 2> $errfile
done
fi

echo "Done!"