Inconsistency of allele depth in BAM and VCF

Question

[CY]

We checked several alleles (locus) in both bam and vcf file. Almost all of them show a little bit high depth in bam file and a bit lower depth (AD) in vcf file. I guess this have something to do with variant caller filtering out some reads based on pre-defined criteria before make the variant calling.

Am I guessing right? If it is, what criteria probably are is this case? Base quality or something else?

[finswimmer]

Hello,

the differences might come from how the variant caller works and how you calculate the read depth in the bam file. You have to tell us what variant caller you use and how do you calculate read depth. Than one can say more.

fin swimmer

[CY]

I used Metect basically with default parameters.

Answer 1

[Noushin N]

You are right about the likely reason of this discrepancy. Essentially, you have to compare the filters used in the utility you are using for bam review (e.g. samtools) and the variant caller.

A few choices are:

• duplicate read exclusion
• read mapping quality score filter
• read passing quality controls
• being part of primary alignment