Mining of Differential Correlation

• Activity Number: 309
• Type: Contributed
• Date/Time: Tuesday, August 11, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Learning and Data Mining
• Author(s): Kelly Nicole Bodwin* and Andrew Nobel and Kai Zhang
• Companies: The University of North Carolina at Chapel Hill and The University of North Carolina at Chapel Hill and The University of North Carolina at Chapel Hill
• Keywords: data mining ; correlation ; genomics ; differential coexpression ; networks ; high dimensional data

Abstract

Given data from two classes of samples, it is often of interest to identify predictors that behave differently in one class than the other. A common approach is to study first-order differences; that is, to discover predictors whose mean values are statistically higher in samples from the first class than in those from the second. Herein we propose a new second-order analysis, Mining of Differential Correlation (MDC). The MDC procedure seeks to identify a group of predictors whose average pairwise correlation amongst samples in the first class is higher than amongst samples in the second class. We employ an iterative procedure that adaptively updates the size and elements of an initial predictor set. These updates are based upon multiple testing of the differential correlation of individual predictors. To perform the tests, we suggest a carefully chosen test statistic and derive its asymptotic distribution under certain mild assumptions. We investigate the performance of MDC by applying it to simulated data as well as gene expression data taken from The Cancer Genome Atlas and other recent experimental datasets.