1 bp insertion supported by deep sequencing missed by bcftools mpile/call

[carolynzy]

Hello,

I'm using the following command to call variants in my bam file:

bcftools mpileup -f ../GCF_000195955.2_ASM19595v2_genomic.txt \
-Q10 -C 50 -d 500 -m 1 -A -a AD,ADF,ADR,DP,SP,INFO/AD,INFO/ADR \
-O v --threads 5 Rv26_sorted.bam | bcftools call -m -M --ploidy 1  -A \
-a GQ,GP -  > Rv26.vcf.gz

The bam file is produced by:

bwa mem -t5 GCF_000195955.2_ASM19595v2_genomic.txt Rv26_dedupe.fastq |\
 samtools view -b - |samtools sort -m 3G -o Rv26_sorted.bam \
--reference GCF_000195955.2_ASM19595v2_genomic.txt --threads 5 \
--output-fmt BAM -

However, I found several indels are missed. For example, between 3590686 and 3590687, there is an insertion of 1 bp. The sequencing depth at this site is >100. If I use IgV to visulize this site, the insertion is supported by > 99% of the reads at that site. It should be called as I think the signal is quite strong. The CIGARs of reads aligned to the region also show this insertion exists.

I shared the bam file and the reference genome here
GCF_000195955.2_ASM19595v2_genomic.txt
& https://1drv.ms/u/s!Ahu3aHGoa85BjYhMCjKwKu5buNjAdQ?e=gqgqHw.

Would you please help me to identify what the problem is? Thank you very much！

Yang