One good idea is to use a job array that takes a samplesheet in input. Each line specifies one sample. Then using SLURM_ARRAY_TASK_ID variable in the slurm bash script you can acces the respective line in your sample sheet and execute a bwa mem instance with the number of CPU specified in the slurm parameters.

something like for 5 CPU - 20G RAM and 10 samples. You may want to tune the size of the RAM required depending on the reference genome.

#!/bin/bash
#SBATCH --nodes=1 --ntasks-per-node=5 --mem-per-cpu=20G
#SBATCH --array=1-10
#SBATCH -J jobName


samplesheet="sample_sheet.txt"

threads=$SLURM_JOB_CPUS_PER_NODE

r1=`sed -n "$SLURM_ARRAY_TASK_ID"p $samplesheet |  awk '{print $1}'` # fastq.gz file

bwa mem -t $threads ref.fa $r1

very simple sample_sheet.txt :

sample_1_R1.fastq.gz
sample_2_R1.fastq.gz
sample_3_R1.fastq.gz
sample_4_R1.fastq.gz

You may change it something more detailed. Here for example it's paired-end data. And column 1 contains the name of the sample.

sample_1    sample_1_R1.fastq.gz    sample_1_R2.fastq.gz
sample_2    sample_2_R1.fastq.gz    sample_2_R2.fastq.gz
sample_3    sample_3_R1.fastq.gz    sample_3_R2.fastq.gz
sample_4    sample_4_R1.fastq.gz    sample_4_R2.fastq.gz

Then within the script you can change the section :

r1=`sed -n "$SLURM_ARRAY_TASK_ID"p $samplesheet |  awk '{print $1}'` # fastq.gz file

to

name=`sed -n "$SLURM_ARRAY_TASK_ID"p $samplesheet |  awk '{print $1}'` 
r1=`sed -n "$SLURM_ARRAY_TASK_ID"p $samplesheet |  awk '{print $2}'` 
r2=`sed -n "$SLURM_ARRAY_TASK_ID"p $samplesheet |  awk '{print $3}'`