Online Program Home
My Program

Abstract Details

Activity Number: 575 - Statistical Methods for Batch Effect Correction and Cell Type Deconvolution
Type: Contributed
Date/Time: Wednesday, July 31, 2019 : 2:00 PM to 3:50 PM
Sponsor: Section on Statistics in Genomics and Genetics
Abstract #307097
Title: Surrogate Variable Analysis Based Deconvolution of Transcriptomics Data
Author(s): Li Dong* and Xiaojing Zheng and Fei Zou
Companies: University of North Carolina at Chapel Hill and University of North Carolina at Chapel Hill and University of North Carolina at Chapel Hill
Keywords: Bulk transcriptomics data; Deconvolution; Cell type proportions; Surrogate variable analysis

Deconvolution of bulk transcriptomics data from mixed cell populations is important to identify cellular mechanism of diseases. Current deconvolution methods either use no prior information or require gene signatures or markers of each cell type which are difficult to obtain. To capitalize data with partial cell proportions available in a fraction of samples, we propose a supervised surrogate variable analysis based deconvolution approach where unmeasured cell proportions are treated as hidden confounding variables and are estimated using relative cell proportions from measured cell populations in a subset of samples. The proposed method takes advantages of whole genome transcriptomics data rather than the existing methods using only selected feature genes. In addition, the proposed method is more robust compared to existing deconvolution methods when other hidden (confounding) variables exist. Simulations demonstrated that our method improves the accuracy of cell proportion estimates. The method was applied on a real dataset of sexually transmitted infection for estimating whole blood cell proportions.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2019 program