vcf_validator expects the wrong number of PL values

[zgoodwin]

Hi there,

I am seeing an issue where the validator complains about the number of values in the PL field being inaccurate, when in fact, the number of PL values at the specified samples and line numbers are consistent with the number of alleles at the site:

Examples from the validator file (vcf_validator_linux):

Error: Sample #355, field PL does not match the meta specification Number=G (expected 6 value(s)). This occurs 1 time(s), first time in line 1304.
Error: Sample #154, field PL does not match the meta specification Number=G (expected 6 value(s)). This occurs 1 time(s), first time in line 1383.

Values in the VCF file for each line/sample combination (only showing the chrom, pos, ID, ref, alt and PL fields for the affected samples):

chr1     203906  .       G       A       0,0,0
chr1     219447  .       C       T       445,54,0

As you can see, the validator seems to think that these sites are tri-allelic (6 PL values) when they're really bi-allelic, therefore there should only be 3 PL values. Is this a bug or could I be doing something wrong?