Cell Type Deconvolution and Transformation ofMicroenvironment Microarray Data

[摘要] Transformations are an important aspect of data analysis. In this work we explore the impact of data transformation on the analysis of high-throughput -omics data. Specifically, we explore two applications where data transformation plays an important role. The first application is estimating cell types using gene expression data. Here we develop dtangle, a method that carefully considers scale transformations when estimating cell type proportion estimates. This method broadly out-performs existing deconvolution methods in a comprehensive meta-analysis. Secondly, we explore the role of simple data transformations for the analysis of microenvironment microarray data. In this section we look at simple data transformations and how they interact with visualization, discovery of latent effects, and data integration. We find that simple transformations applied alone or in sequence can make salient important aspects of the data.

[发布机构] University of Michigan

[效力级别] Statistics and Numeric Data

[关键词] Cell Type Deconvolution and Transformation of Microenvironment Microarray Data;Statistics and Numeric Data;Science;Statistics