Performance improvement with protocol optimization; array sensitivity and specificity with application of sPROFILER to GSEQ calls. (data shown for Cincinnati arrays). Data is arranged in the same patient ID order as figure 1. (a) False positive calls with and without protocol optimization/sPROFILER. No-calls and positive calls were processed for the first 12 chips (short and long range PCR protocol) while only no-calls were processed for the remaining 13 chips (short range only PCR protocol). No-calls were converted to wild-type, left as no-call, or were assigned a variant call. Chips that were analyzed only for no-calls may show an increase in false positive rate due to conversion of a fraction of no-calls to variant calls, some of which are not true variants. (b) False negative calls with and without protocol optimization/sPROFILER represented as a portion of total true variants.