Removing Unwanted Variation from High-Dimensional Data with Negative Controls

• Activity Number: 156
• Type: Invited
• Date/Time: Monday, August 5, 2013 : 10:30 AM to 12:20 PM
• Sponsor: WNAR
• Author(s): Johann Gagnon-Bartsch*+ and Laurent Jacob and Terence Speed
• Companies: Berkeley and Berkeley and The Walter & Eliza Hall Institute of Medical Research
• Keywords: negative controls ; microarrays ; unwanted variation ; differential expression

Abstract

High dimensional biological data suffer from unwanted variation, such as the batch effects common in microarray data. Unwanted variation complicates the analysis of high dimensional data. For example, in DE analyses, unwanted variation may lead to high rates of false discoveries, high rates of missed discoveries, or both. In many cases the factors causing the unwanted variation are unknown and must be inferred from the data. In such cases, negative controls may be used to identify the unwanted variation and separate it from the wanted (biological) variation. The exact method by which this can be accomplished depends critically on the final goal of the analysis, e.g. DE, classification, or clustering. In a paper published in 2012 in Biostatistics, we presented a method called RUV-2 for use in DE analyses. In this talk I will briefly review RUV-2. I will then describe a new method, RUV-4, that often outperforms RUV-2. RUV-4 is relatively insensitive to the number of unwanted factors included in the model; this makes estimating the number of factors less critical. Finally, I will discuss similar methods that can be used for classification and clustering analyses.