Return
Robust Spatial Cell-Type Deconvolution with Qualitative Reference for Spatial Transcriptomics
DOI:10.1002/smtd.202401145.png)
Abstract
En 中文
Many spatially resolved transcriptomic technologies have been developed to provide gene expression profiles for spots that may contain heterogeneous mixtures of cells. To decompose cellular composition and expression levels, various deconvolution methods have been developed using single-cell RNA sequencing (scRNA-seq) data with known cell-type labels as a reference. However, in the absence of a reliable reference dataset or in the presence of heterogeneous batch effects, these methods may introduce bias. Here, a Qualitative-Reference-based Spatially-Informed Deconvolution method (QR-SIDE) is developed for multi-cellular spatial transcriptomic data. Uniquely, QR-SIDE provides a detailed map of spatial heterogeneity for individual marker genes and performs robust deconvolution by adaptively adjusting the contributions of each marker gene. Simultaneously, QR-SIDE unifies cell-type deconvolution with spatial clustering and incorporates spatial information via a Potts model to promote spatial continuity. The identified spatial domains represent a meaningful biological effect in potential tissue segments. Using simulated data and three real spatial transcriptomic datasets from the 10x Visium and ST platforms, QR-SIDE demonstrates improved accuracy and robustness in cell-type deconvolution and its superiority over established methods in recognizing and delineating spatial structures within a given context. These results can facilitate a range of downstream analyses and provide a refined understanding of cellular heterogeneity.
Keywords:
cell-type deconvolution
mixture Poisson regression model
Poisson factor model
spatial transcriptomics
Journal
IF:
9.1
Papers:
4.4K
Citations:
2.8W
Organization
No organization information available
Cited Papers
CytoMAP: A Spatial Analysis Toolbox Reveals Features of Myeloid Cell Organization in Lymphoid Tissues
CELL REPORTS
IF6.9
Bulk tissue cell type deconvolution with multi-subject single-cell expression reference
NATURE COMMUNICATIONS
IF15.7
Integration of spatial and single-cell transcriptomic data elucidates mouse organogenesis
NATURE BIOTECHNOLOGY
IF41.7

