Online Adjustment for Unwanted Variation in Molecular Assays

• Activity Number: 470
• Type: Topic Contributed
• Date/Time: Wednesday, August 7, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Terence Speed*+ and Johann Gagnon-Bartsch and Laurent Jacob
• Companies: The Walter & Eliza Hall Institute of Medical Research and Department of Statistics, University of California at Berkeley and Berkeley
• Keywords: Molecular assay ; Miroarray gene expression ; Random effects models ; Personalized medicine ; Differential expression of molecular species ; On-line adjustment

Abstract

In this talk I shall assume that the version of personalized medicine under consideration involves the use of a molecular assay on a sample from a relevant tissue, say a microarray gene expression profile on a tumor biopsy. Such assays are prone to being affected by a variety of forms of unwanted variation, such as those due to sample quality, sample processing and storage, and assay reagents, operators, equipment and environmental conditions. As a result, steps will usually be taken at the time the assay results are analyzed and interpreted, to adjust for or remove unwanted variation. We have recently discussed a variety of methods of doing this in the context of differential expression of molecular species, methods which are equivalent to the use of certain random effects models. In this talk I will briefly describe these methods, and then outline our efforts to adapt them to the situation most likely to be met in the context of personalized medicine, where samples are assayed individually or in batches, and the objective is to assign them to predefined categories on the basis of the molecular measurements made on them.