I think your best bet is to contact the company. Few people on this board use this tool. When you do contact them make sure to specify what your polyplidity consist of, the term can describe many alternative scenarios.

I also think that you are probably seeking something like the maximum expected variation, or perhaps a probability cutoff. My feeling is that a quantity such as minimum variant frequency should be determined primarily from the quality and coverage of your data and not as an external parameter. I could be wrong though.