What percentage of a sample’s genome needs to be diploid for accuracy? Is there a limit to the number of CNVs in my sample for the algorithm to be accurate?

The biggest constraint to consider is the fraction of the amplicons that have CNVs. Currently, the pipeline assumes that the median ploidy is 2. If a sample is used such that most of the amplicons (>=50%) have a CNV, the automated report will give inaccurate results, and the sample should be analyzed manually.