In answer to my question, "Would you share the filter parameters used at YFull to separate Novel SNPs into Best qual, Acceptable qual and Ambiguous qual groups?
Their response: "Not all of them:
Best qual: 5 reads and more, 100% quality.
Acceptable qual: 3 reads and more, 91% quality and more.
Ambiguous qual: 2 reads and more, 85% quality and more.
Quality means percent of right value in all reads of this position.
And we use some heuristics for regions also."

Now a question remains; Why is there multiple "Reads"? One of my "Best Novel SNPs" showed T for all 141 Reads (Ref = C).