Francesco Vallania
Postdoctoral Scholar
Khatri Lab, Stanford University

Abstract:
Cell-mixture deconvolution is an in silico strategy that quantifies proportions of multiple cell subsets directly from gene expression of mixed-cell samples using a signature matrix to link expression to cell types. Despite its accuracy, deconvolution is significantly affected by the platform used to measure expression and by the disease state of the sample, severely limiting its applicability. Here, we present “immunoStates”, a new blood cell signature matrix designed to remove technical and biological biases to extend deconvolution to all available data.