VCF File Columns

When a VCF file is exported by exporting sets or exporting table rows, ArrayStar adheres to the Variant Call File (VCF) v. 4.2 specification:

 

      In the FILTER field, each row is marked with one of three qualifiers to show whether or not a position was covered:

 

“PASS” for positions where a call could be determined based on the sequence read data.

 

“NC” for positions with no sequence read coverage (this will be denoted at the top of the file under ##FILTER.)

 

“.” for positions when data for a call is missing or a call could not be made.

 

These changes to the FILTER field apply to both single-sample and multi-sample VCFs, but not to VCFs lacking any sample information.

 

      In the QUAL field, a Phred-scaled quality score is provided for the assertion made in the ALT column. The score is calculated as -10 log10 prob (call in ALT is wrong).

 

In rows where the ALT column contains ‘.’ (i.e. no variant was called), the column contains -10log10 prob(variant).

 

In rows where the ALT does not contain ‘.’ (i.e. a variant call), the column contains -10log10 prob(no variant).

 

A missing value is specified as “.”

 

      The PA field contains the Pnotref value. Note that the QUAL scale is reversed relative to Pnotref when ALT is "."; that is, when a position is in the reference. However, in one direction or the other, it will scale logarithmically with Pnotref. This does mean that it will be closer to Qcall (or "GQ") in cases where there isn't “homozygous vs. heterozygous” call ambiguity. However, when the ambiguity is present, it will diverge.