In my merge_output.vcf.gz file most of the time I'm getting just a single variant per position. However I also get one position where there are two variants, but their allele frequency does no add up to 1, nor do the allelic depths add up to the total read depth for the position. Am i misunderstanding the values or is there something wrong happening with the calculations?

A snippet of my merge_output.vcf.gz looks like this:

#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO	FORMAT	SAMPLE
K03455.1	2229	.	C	A	22.46	PASS	F	GT:GQ:DP:AD:AF	1/1:22:19666:255,18257:0.9284
K03455.1	2259	.	G	A	36.43	PASS	F	GT:GQ:DP:AD:AF	1/1:36:19721:182,18951:0.9610
K03455.1	3161	.	G	A,T	18.93	PASS	F	GT:GQ:DP:AD:AF	1/2:18:20313:190,10274,17655:0.5058,0.8691

Where the last row is the one with multiple variants.

Any help would be greatly appreciated