A1020
Title: SDePER: A hybrid machine learning and regression method to deconvolve spatial barcoding-based transcriptomic data
Authors: Yunqing Liu - Yale School of Public Health (United States)
Ningshan Li - Yale School of Public Health (United States)
Ji Qi - Yale School of Public Health (United States)
Gang Xu - Yale School of Public Health (United States)
Jiayi Zhao - Yale School of Public Health (United States)
Nating Wang - Yale School of Public Health (United States)
Xiayuan Huang - Yale School of Public Health (United States)
Wenhao Jiang - Yale School of Public Health (United States)
Aurelien Justet - Yale School of Medicine (United States)
Taylor Adams - Yale School of Medicine (United States)
Robert Homer - Yale School of Medicine (United States)
Amei Amei - University of Nevada (United States)
Ivan Rosas - Baylor College of Medicine (United States)
Naftali Kaminski - Yale School of Medicine (United States)
Zuoheng Wang - Yale University (United States)
Xiting Yan - Yale School of Medicine (United States) [presenting]
Abstract: Spatial barcoding-based transcriptomic (ST) data require cell-type deconvolution for cellular-level downstream analysis. The aim is to present SDePER, a hybrid machine learning and regression method, to deconvolve ST data using reference single-cell RNA sequencing (scRNA-seq) data. SDePER uses a machine learning approach to remove the systematic difference between ST and scRNA-seq data (platform effects) explicitly and efficiently to ensure the linear relationship between ST data and cell type-specific expression profile. It also considers the sparsity of cell types per capture spot and across-spots spatial correlation in cell type compositions. Based on the estimated cell-type proportions, SDePER imputes cell-type compositions and gene expression at unmeasured locations in a tissue map with enhanced resolution. Applications to coarse-grained simulated data and four real datasets showed that SDePER achieved more accurate and robust results than existing methods, suggesting the importance of considering platform effects, sparsity and spatial correlation in cell-type deconvolution.
