Emerging advances in genomic sequencing have prompted the development of new computational methods for studying the genomic sources of human diseases. This paper presents a recent statistical approach for detection of local regions with significant copy number alterations (CNAs) in lung cancer population. Mapping such regions is of interest as they are potentially associated with lung cancer. Conventional application of multiple testing methods corresponds to testing for CNAs at each probe separately and thresholding the t-statistics as test statistics. Due to the large number of probes, this approach often fails to detect CNA regions. In contrast, the proposed method uses the heights of located peaks and improves the detection power. This is achieved by taking advantage of the spatial structure in the data as well as reducing the number of tests in the multiple comparisons problem. In copy number analysis, it is common to apply segmentation or change detection tools to each individual genomic sample. However, since segmentation results vary among subjects, it becomes difficult to find the common genomic regions in population analyses. Our approach solves this problem by performing the analysis using summary statistics to study at population level directly. Hence, the region detection is performed on the summary t-statistic map. The proposed method is applied to lung cancer data and shows promise for detection of local regions with significant CNAs.
Keywords: Copy number alterations (CNAs); Gene-Cancer association; Local region detection; Lung cancer.