Hello, I've recently encountered a situation in which a SNV overlaps a single base pair deletion variant. The output from GATK v3.6 HaplotypeCaller is as follows:
8 145738767 . CG C 7919.73 . DP=619;MLEAC=1;MLEAF=0.500;QD=12.92 GT:AD:DP:GQ:PGT:PID:PL:SAC 0/1:305,308:613:99:1|0:145738764_AC_A:7957,0,8498:151,154,143,165
8 145738768 . G T 8454.77 . DP=615;MLEAC=1;MLEAF=0.500;QD=13.77 GT:AD:DP:GQ:PGT:PID:PL:SAC 0/1:309,305:614:99:0|1:145738764_AC_A:8483,0,9180:143,166,151,154
It is very difficult to mutually interpret these records because they overlap in their reference range. For example, the second records suggests that there is indeed a reference G allele at 8:145738768, and that it's depth is 309. Similarly, the second record suggests there is a reference CG allele, and that it's depth is 305.
I suspect this is one example amongst a large class of "reference overlapping" variants that HC might call. Is there a more subtle interpretation of VCF that would make this clear? At present, this is impacting some downstream VCF processing. I am happy to provide a minimal BAM if it would be helpful.
Many thanks!