Motivation: Cancer genomes are characterized by the accumulation of point mutations and structural alterations such as copy-number alterations and genomic rearrangements. Among structural changes, systematic analyses of copy-number alterations have provided deeper insight into the architecture of cancer genomes and had led to new potential treatment opportunities. During the course of cancer genome evolution, selection mechanisms are leading to a non-random pattern of mutational events contributing to fitness benefits of the cancer cells. We therefore developed a new method to dissect random from non-random patterns in copy-number data and thereby to assess significantly enriched somatic copy-number aberrations across a set of tumor specimens or cell lines. In contrast to existing approaches, the method is invariant to any strictly monotonous transformation of the input data which results to an insensitivity of differences in tumor purity, array saturation effects and copy-number baseline levels.
Results: We applied our approach to recently published datasets of small-cell lung cancer and squamous cell lung cancer and validated its performance by comparing the results to an orthogonal approach. In addition, we found a new deletion peak containing the HLA-A gene in squamous cell lung cancer.