Number of allelic depths is larger than number of inferred alleles in VCF file

Question

[Alexandros]

Hello,

I performed variant calling and genotyping of my samples using GATK 4.2.6.1. What I found in the VCF file is that despite having only 2 genotypes separated with "/" or "|", the number of allelic depths (ADs) is larger than the number of alleles (i.e., in the example below 3 ADs for 2 alleles or 4 ADs for 1 allele in the second example). What is the reason for this?

1|2:11,70,65:146:99:0|1:2857836_C_*:5031,2213,2478,2377,0,2643:2857836
3|3:0,0,0,4:4:13:1|1:1933943_TTAAGGTAG_T:139,139,139,139,139,139,13,13,13,0:1933943

I thought that in this case only the depths for alleles 1 and 2 (or for the allele 3 in the second example should be given). Do the rest of ADs correspond to the other alleles for example of the reference or of other samples?

cheers

[Jeremy Leipzig]

show us the whole line of this VCF, at least CHROM POS ID REF ALT QUAL FILTER INFO FORMAT

[Alexandros]

##source=CombineGVCFs

##source=SelectVariants

#CHROM  POS ID  REF ALT QUAL    FILTER  INFO    FORMAT  Bc_ref_BXQ_D-illumina_reads Bc_ref_BXQ_E-illumina_reads

contig_17   2857838 .   C   *,A 2203.81 .   AC=1,1;AF=0.500,0.500;AN=2;BaseQRankSum=0.628;DP=397;ExcessHet=0.0000;FS=6.717;MLEAC=1,1;MLEAF=0.500,0.500;MQ=59.73;MQRankSum=3.53;QD=15.09;ReadPosRankSum=3.22;SOR=0.185   GT:AD:DP:GQ:PGT:PID:PL:PS   1|2:11,70,65:146:99:0|1:2857836_C_*:5031,2213,2478,2377,0,2643:2857836  ./.:232,0,0:232:0:.:.:0,0,0,0,0,0

[Alexandros]

And for the second genotype above

##source=CombineGVCFs

##source=SelectVariants

#CHROM POS ID REF ALT QUAL FILTER INFO FORMAT Bc_ref_BXQ_D-illumina_reads Bc_ref_BXQ_E-illumina_reads

contig_9    1933948 .   G   T,C,*   234.21  .   AC=1,1,2;AF=0.250,0.250,0.500;AN=4;DP=17;ExcessHet=0.0000;FS=0.000;MLEAC=1,1,2;MLEAF=0.250,0.250,0.500;MQ=45.91;QD=23.42;SOR=3.258  GT:AD:DP:GQ:PGT:PID:PL:PS   1/2:0,2,4,0:6:72:.:.:252,168,162,84,0,72,252,168,84,252 3|3:0,0,0,4:4:13:1|1:1933943_TTAAGGTAG_T:139,139,139,139,139,139,13,13,13,0:1933943

[dthorbur]

I agree with @leipzig. For this position there are 4 alleles total: 1 reference (G), and 3 alternatives (T,C, *; where * is overlapping a deletion in GATK), so the AD annotation describes the number of unfiltered reads that support each allele in the order reported.

Here, in sample Bc_ref_BXQ_D-illumina_reads at position contig_9:1933948 you have 0 reads supporting G, 2 supporting T, 4 reads supporting C, and 0 reads supporting a deletion.

Answer 1

[Jeremy Leipzig]

So * is the GATK-specific spanning or overlapping deletion. It's just another allele here, so if you have

G -> T / C / *

I would expect 4 AD's and

C -> * / A

I would expect 3 AD's