Optimizing KMCP runs (SLURM batch file)

Input stats

file                              format  type    num_seqs        sum_len  min_len  avg_len  max_len
DRR171473_1.fastp.depleted.fq.gz  FASTQ   DNA   20,312,778  2,754,850,269       28    135.6      150
DRR171473_2.fastp.depleted.fq.gz  FASTQ   DNA   20,312,778  2,745,109,847       26    135.1      150

kmcp-single.sh

#!/bin/sh
#SBATCH --partition=general
#SBATCH --qos=medium
#SBATCH --time=24:00:00
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=32
#SBATCH --mem=100GB
#SBATCH --gres=jobspernode:one:1
#SBATCH --mail-type=END

set -xeuo pipefail
# perhaps with the path included
accession='DRR171473'
source_folder="/path/to/source/folder"
DB_INDEX="/path/to/humgut.index/database.index"
THREADS=64
WORK_DIR="/dev/shm/$USER/kmcp"
LOG_DIR="/path/to/source/folder"
TAXID_MAP=/path/to/humgut_db/taxid-gtdb.map
TAXDUMP=/path/to/humgut_db/taxdump
DB_NAME="kmcp_humgut"
mkdir -p "$LOG_DIR"
mkdir -p "$WORK_DIR"
cd "$WORK_DIR"
echo "copying source files"
cp "$source_folder/${accession}_1.fastp.depleted.fq.gz" .
cp "$source_folder/${accession}_2.fastp.depleted.fq.gz" .
echo "execusting parallel search"
# execute the kmcp search with parallel
THREADS_PER_JOB=12
cat "$DB_INDEX" | parallel -j $(($THREADS / $THREADS_PER_JOB)) kmcp search \
    --db-dir {} \
    --threads $THREADS_PER_JOB \
    --load-whole-db \
    "${accession}_2.fastp.depleted.fq.gz" \
    "${accession}_2.fastp.depleted.fq.gz" \
    -o "${accession}.${DB_NAME}@{#}.tsv.gz" \
    --log "${LOG_DIR}/${accession}.${DB_NAME}@{#}.log"

THREADS=64
kmcp merge ${accession}.${DB_NAME}@*.tsv.gz \
    -o "${accession}.${DB_NAME}.tsv.gz" \
    --log "${LOG_DIR}/${accession}.${DB_NAME}.log"

kmcp profile --taxid-map "$TAXID_MAP" --taxdump "$TAXDUMP" \
    "${accession}.${DB_NAME}.tsv.gz" \
    -m 3 -j $THREADS \
    -o "${accession}.${DB_NAME}.profile" \
    --metaphlan-report "${accession}.${DB_NAME}.profile.metaphlan" \
    --log "${LOG_DIR}/${accession}.${DB_NAME}.profile.log"