The methods above, in conjunction with a map of common polymorphisms (both single nucleotide and copy number) and hybrid arrays for detection6, allow characterization of the genetic variation in each sample with high accuracy and in a more comprehensive manner than previously possible. In addition to providing discrete calls, this framework provides a confidence of each call, which serves as a good guideline as to data and genotype quality. (In the downstream analyses that follow, we use these confidences only as a threshold for inclusion or exclusion of data; we note that such filtering has the potential to introduce bias, and methods that incorporate the uncertainty may perform better.)
