Identification of High-Risk Cells in Single-Cell Spatially Resolved Transcriptomics Data Using Deep Transfer Learning
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
The examination of high-risk cells and regions in tissue samples from spatially resolved transcriptomics platforms, offers meaningful insights into specific disease processes. For existing methods, while cell types or clusters can be identified and associated with disease attributes, individual cells are unable to be associated in the same manner which may result in failing to identify subsets of cells associated with disease attributes especially if the disease-associated cells cluster together with non-disease-associated cells.
Method
DEGAS (Diagnostic Evidence Gauge of Single-cells) [5], solves the above problem with a sophisticated deep transfer learning algorithm designed to identify high-risk components in single-cell RNA sequencing data from tumor samples. DEGAS employs latent representations of gene expression data and domain adaptation to transfer disease attributes from patients to individual cells. In this research, we present DEGAS’s versatility in adapting to data arising from single cell spatially resolved transcriptomics platforms such as the 10X Genomics Xenium platforms, and Nanostring’s CosMx platform. By integrating spatial location information from the above platforms, DEGAS can not only identify high-risk components in tissue samples but also pinpoint locations within the slides associated with disease status.
Results
We evaluated DEGAS across multiple platforms, including 10X Genomics Xenium and Nanostring CosMx. DEGAS successfully identified high-risk cells and regions, which were validated through known markers. Additionally, DEGAS was applied to our newly generated T2D Xenium dataset and a publicly available melanoma Xenium dataset. We tested DEGAS on publicly available Nanostring CosMx FFPE samples of normal and Hepatocellular Carcinoma tissues, revealing high-risk cells and topologies associated with key pathways. Notably, high-risk regions were predominantly enriched in tumor tissue, with DEGAS uncovering heterogeneity that correlates with aggressive disease markers and cell type diversity.
