PSEA: Population-Specific Expression Analysis

The characterization of molecular changes in diseased tissues can provide crucial information about pathophysiological mechanisms and is important for the development of targeted drugs and therapies. However, many disease processes are accompanied by changes of cell populations due to cell migration, proliferation or death. Identification of key molecular events can thus be overshadowed by confounding changes in tissue composition. To address the issue of confounding between cell population composition and cellular expression changes, we developed Population-Specific Expression Analysis (PSEA) [1, 2]. This method works by exploiting linear regression modeling of queried expression levels to the abundance of each cell population. Since a direct measure of population size is often unobtainable (e.g. from human clinical or autopsy samples), PSEA instead tracks relative cell population size via levels of mRNAs expressed in a single population only. Thus, a reference measure is constructed for each cell population by averaging expression data for cell-type-specific mRNAs derived from the same expression profile. Here we will demonstrate some of the functionalities in the PSEA package. We will first generate reference signals and deconvolve individual transcripts to illustrate the method. We will then show how to apply PSEA to entire expression profiles. Let us start by loading the package > library(PSEA) We have included expression data obtained from brain samples of 41 individuals as well as their phenotypes, i.e. control and Huntington’s disease (HD) (the full data is deposited at http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE3790) > data(example) The example data contains the variable expression, a matrix with the expression levels of 23 transcripts and the variable groups, a vector with phenotypic information encoded as 0 and 1 (indicating control and disease, respectively). Detailed information about the data is provided in the corresponding manual pages (see ?expression and ?groups).