Hotel Estoril Eden, Monte Estoril,
5-8 October 2005



NextText Box: Participants
Text Box: Programme

spects of Bayesian Gene eXpression (BGX): Assessment of
Uncertainty and Inference Without Replicates

Anne-Mette K. Hein and Sylvia Richardson
Imperial College London, UK

BGX (Bayesian gene expression, Hein et al, 2005, Biostatistics) is an integrated approach to the analysis of A ymetrix GeneChip arrays. The approach relies on a Bayesian hierarchical model for probe level GeneChip data. Background correction, gene expression level estimation and assessment of differential expression are performed simultaneously. Full posterior distributions of the model parameters can be obtained through MCMC techniques. We study here two aspects of the BGX approach that in particular distinguishes it from other approaches: the integration of di erent steps in the analysis of A ymetrix GeneChips that are usually performed in separate steps, and the estimation of posterior distributions of gene expression levels rather than point estimates. We study implications of these features in relation to the estimation of uncertainty measures and the possibility of performing inference when no replicate arrays are available.