A new method to assess the statistical significance of copy number changes identified in genomic DNA by array comparative genome

The importance of cytogenetically visible rearrangements in human genetic disease has long been recognised and there is now abundant evidence showing that smaller, less readily detectable chromosomal rearrangements can also be clinically important. The full significance and extent to which such cryptic rearrangements contribute to human genetic disease has yet to be determined. One way of elucidating this is by comparative genome hybridization (CGH) to DNA microarrays (array CGH). However, recent array CGH studies noting false negative and false positive results raise important concerns regarding the suitability of the approach for the detection of constitutional chromosomal rearrangements, particularly in a clinical diagnostic environment where a robust assay, providing clear, high quality results of measurable significance is required. Here, we present the results of array CGH studies that address these concerns. Using tiling path arrays for the terminal 2Mb of chromosome 16p and telomere specific arrays we show that probes vary in their usefulness for detecting copy number changes and that sequence composition cannot be used to predict the probe behaviour. Significantly, we present a powerful new method of statistical analysis that overcomes problems of probe specificity and sensitivity by combining fluorescence signals from arrayed probes that map to neighbouring genomic locations. The method uses a modification of the Smith-Waterman algorithm that also provides a measure of robustness. We tested the method using hybridisation data from the 2Mb 16p tiling path array and show that 100% of mo nosomies ›250kb were identified with a high degree of significance and that the boundaries of the deletions were accurately and robustly located. The studies are an important step towards optimising array CGH based approaches and improving their suitability for clinical diagnostic purposes.