| Analysis | Data from the focused confirmation array were analysed using the ADM-2 algorithm at threshold 4. The ADM-2 algorithm uses log2 ratios weighted by log2 ratio error as calculated by Feature Extraction software to identify genomic intervals with copy number differences between the sample and the reference. Data were centralized, and calls with average log2 ratios less than 0.3 were excluded from the analysis, as were any calls detected by probes containing a known SNP that may alter an AluI or RsaI restriction site as determined by the 9.3 million in the UCSC annotation database for the genome browser (Karolchik D. et al. The UCSC Genome Browser Database. Nucleic Acids Res. (2003) 31:51-54.). The false-positive rate for the ADM-2 algorithm at threshold 4 was determined using three self-self hybridizations of the reference sample. In the three replicate self-self experiments, ADM-2 analysis with threshold 4 identified an average of 29.6 single-probe intervals and 10.3 multi-probe intervals. Comparing the average number of variant interval calls in self-self experiments with the average number of variant interval calls for each sample, we estimated the false-positive rate to be 0.05 (10.3/197) for multi-probe calls and 0.04 (29.6/669) for single-probe calls. The false-negative rate was estimated in a manner similar to that described by Wong et al. (Wong et al. A comprehensive analysis of common copy-number variations in the human genome. Am. J. Hum. Genet. (2007) 80:91-104.) based on four replicate experiments for one of the samples. In four replicate experiments, 223 putative variant intervals were observed two or more times and were considered true calls (49 intervals were observed twice, 43 intervals were observed three times and 131 intervals were observed four times), yielding an estimate of false-negative rate of 0.16 [(2*49+43)/(4*223)=0.16]. In this analysis, we conservatively considered aberrant intervals in two experiments the same if they overlapped by more than 0.9. |
