Abstract:

I will discuss a general framework for detection of local signals, primarily defined by change-points, in sequences of data, with a focus on applications in genomics. Changes can occur continuously, e.g., a change in the slope of a regression line, or discontinuously, e.g., a jump in the level of a process. A motivating example of jump discontinuities is provided by copy number variation (CNV). Data can be based on Comparative Genomic Hybridization, Single Nucleotide Polymorphisms (SNPs) or DNA resequencing. For the first two it is often plausible to assume that the data are normally distributed, and for the third that the data form a Poisson random field. After discussing problem formulation and appropriate models, I will focus on the simplest version of the problem: segmentation of independent normal observations according to abrupt changes in the mean. Results will be illustrated by simulations and by applications to the BT474 cell line. To illustrate the generality of the methods, I will also discuss changes in slope of a regression line and applications. Some difficulties associated with dependent observations will be mentioned, and if there is enough time confidence regions for the change-points and joint regions for the change-points and mean values will also be discussed.