An Empirical Evaluation of Array Normalization for Agilent MicroRNA Expression Arrays

• Activity Number: 359
• Type: Contributed
• Date/Time: Tuesday, August 2, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Li-Xuan Qin*+ and Jaya M. Satagopan and Sam Singer
• Companies: Memorial Sloan-Kettering Cancer Center and Memorial Sloan-Kettering Cancer Center and Memorial Sloan-Kettering Cancer Center
• Keywords: microarray ; microRNA ; normalization

Abstract

Methods for array normalization have been developed for mRNA expression arrays, such as median normalization and quantile normalization. These methods assume few or symmetric differential expression of markers on the array. The performance of the existing normalization methods need to be re-evaluated when applied to microRNA arrays, which consist of a few hundred markers and a reasonable fraction of them are anticipated to have disease-relevance. We empirically examined sources of variations in miRNA array data using a set of Agilent arrays in liposarcoma (n=56) and evaluated normalization methods using Solexa sequence data on a subset of these tumors (n=29) as the gold standard. We found that there is minimum variation between replicate probes for the same target sequence and moderate variation between multiple target sequences for the same miRNA. There is moderately high correlation between Agilent data and Solexa data. Quantile normalization has slightly improved the correlation with Solexa data, as well as the detection of differentially expressed microRNAs both in terms of statistical significance and the direction of change.