Statistical Estimation of Genomic Tumoral Alterations

Yi Liu 1, 2 Christine Keribin 1, 2 Tatiana Popova 3 Yves Rozenholc 2
2 SELECT - Model selection in statistical learning
Inria Saclay - Ile de France, LMO - Laboratoire de Mathématiques d'Orsay, CNRS - Centre National de la Recherche Scientifique : UMR
Abstract : Characterization of the tumoral genomic alterations is an important step in the development of personalized medicine in cancerology. Among the methods for treating micro-array data, the GAP (Genome Alteration Print) method of Popova et al. (2009) characterizes the mutations based on the segmentation of copy number and B-allele frequency signals obtained on each SNP. It uses a deterministic criterion that we propose to replace by a parametric probabilistic model. In this way, we define a Gaussian mixture model whose classes characterize the mutation types. This model is estimated by maximum likelihood through the EM algorithm, allowing us to obtain the estimation of the parameters and the characterization of tumoral alterations on each segment. In our approach, the tumoral ploidy is deduced from a penalized model selection criterion. Our model is tested on simulated data and real data.
Complete list of metadatas

Cited literature [7 references]  Display  Hide  Download

https://hal.inria.fr/hal-01260716
Contributor : Christine Keribin <>
Submitted on : Friday, January 22, 2016 - 3:02:32 PM
Last modification on : Thursday, April 11, 2019 - 4:02:12 PM
Long-term archiving on : Friday, November 11, 2016 - 4:37:16 PM

File

16-GAP-JDS15.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01260716, version 1

Citation

Yi Liu, Christine Keribin, Tatiana Popova, Yves Rozenholc. Statistical Estimation of Genomic Tumoral Alterations. 47èmes Journées de Statistique de la SFdS, Jun 2015, Lille, France. ⟨hal-01260716⟩

Share

Metrics

Record views

540

Files downloads

224