Related Experiment Video
Updated: Sep 14, 2025
09:19
Spatial Profiling of Protein and RNA Expression in Tissue: An Approach to Fine-Tune Virtual Microdissection
Published on: July 6, 2022
5.0K
STHD: probabilistic cell typing of single spots in whole transcriptome spatial data with high definition
Chuhanwen Sun1, Yi Zhang2,3,4,5,6
1Department of Neurosurgery, Duke University, Durham, NC, USA.
Genome Biology
|July 18, 2025
Summary
We developed STHD, a machine learning tool for cell typing in spatial transcriptomics. It accurately identifies cell types in subcellular spots, revealing tissue architecture and immune interactions in tumors.
Area of Science:
- Genomics
- Computational Biology
- Bioinformatics
Background:
- Spatial transcriptomics offers subcellular resolution for gene expression profiling.
- High sparsity and dimensionality pose computational challenges for data analysis.
Purpose of the Study:
- To present STHD, a novel computational method for probabilistic cell typing in high-definition spatial transcriptomics.
- To address the challenges of sparsity and dimensionality in spatial transcriptomics data.
Main Methods:
- Developed a machine learning model (STHD) integrating count statistics and neighbor regularization.
- Applied STHD to analyze whole-transcriptome spatial transcriptomics data at subcellular resolution.
Main Results:
- STHD accurately predicts cell type identities of subcellular spots.
- Revealed global tissue architecture and local multicellular neighborhoods.
- Demonstrated utility in analyzing cell type-specific gene expression and immune interaction hubs within the tumor microenvironment.
Conclusions:
- STHD provides accurate probabilistic cell typing for high-definition spatial transcriptomics.
- The method is generalizable across diverse samples, tissues, and diseases.
- Enables deeper insights into tissue organization and cellular interactions.
