I have a VCF file and I found something strange at a multi-allelic site.

The first few columns are:

 chr12  106633569   rs3088113   C   T,G

The QUAL is high and VQSR says PASS for this site.

For the individual genotypes, there are (GT:AD:DP:GQ:PL):

0/1:34,1,36:71:99:957,1047,1737,0,687,586
0/1:54,3,67:124:99:1662,1760,2981,0,1171,1029
0/2:40,9,0:49:99:136,0,1352,256,1379,1635
1/1:2,1,96:99:99:2896,2895,2921,259,280,0
etc.

From my understanding, 0/1 should have something like 50,50,0 for AD instead of 50,0,50 as in the above individual calls. Also, from this post, I see that the PL field should have an order of AA,AB,BB,AC,BC,CC, where I have 0 PL for AC (i.e. 0/2) for the first individual, but it was called 0/1.

Has anyone seen this kind of problems in his/her VCF file? How did this happen?