Statistical estimation of genomic alterations of tumors

Recent research reveals that personalized medicine is one major way to treat cancer. In order to develop personalized medicine, characterizing the genomic alterations is a vital component. Several methods have been proposed to this end. One of the rst methods is the Genome Alteration Print (GAP) by Popova et al, which uses a deterministic approach. We follow this approach and develop a parametric probabilistic model for GAP, together with its statistical estimation, based on a preliminary segmentation of SNP measurements obtained from microarray experiments. For estimation, we implement the expectation-maximization (EM) algorithm to maximize the likelihood of this model and get the parameter estimation which characterizes the genomic alterations. In our approach, the tumoral ploidy is deduced from penalized model selection. Our model is tested on simulated data and real data