Because the mutation rate to create de novo CNVs is exceedingly low6,16, when observed in individuals with a sporadic disease phenotype they are particularly good candidates to be causal factors5. However, in searching for de novo events it is critical not only to evaluate the evidence in favor of a CNV in a proband (or a tumor), but also to accurately estimate the likelihood against the presence of the CNV in the parents (or normal tissue). Birdseye is designed to address this need: in addition to reporting the evidence in favor of a CNV, it can be used to reassess a discovered region in other samples (such as the proband’s parents) using a framework that does not involve a stringent genome-wide discovery threshold (Fig. 4b,c). Thus one can filter a list of CNVs on the basis of strong evidence against variation in parents, as opposed to simply failing to achieve genome wide-significant evidence in favor of variation (which is frequently a false negative).
