Related Experiment Video
Updated: Jul 1, 2025
10:12
Droplet Barcoding-Based Single Cell Transcriptomics of Adult Mammalian Tissues
Published on: January 10, 2019
18.5K
Adjustment of scRNA-seq data to improve cell-type decomposition of spatial transcriptomics
Lanying Wang1, Yuxuan Hu1, Lin Gao1
1School of Computer Science and Technology, Xidian University, Xi'an 710100, China.
Briefings in Bioinformatics
|March 1, 2024
Summary
Spatial transcriptomics (ST) methods often lack single-cell resolution. We developed a transfer learning framework to adjust single-cell RNA sequencing (scRNA-seq) data, improving cell-type decomposition accuracy in ST data.
Area of Science:
- Computational Biology
- Genomics
- Bioinformatics
Background:
- Sequencing-based spatial transcriptomics (ST) technologies capture gene expression within tissue microenvironments but typically lack single-cell resolution.
- Each ST spot may contain cells from multiple cell types, necessitating computational methods to infer cell type composition.
- Existing cell-type decomposition methods integrate single-cell RNA sequencing (scRNA-seq) data but often overlook distribution differences between scRNA-seq and ST datasets, leading to inaccurate cell-type-specific gene expression profiles.
Purpose of the Study:
- To develop a novel instance-based transfer learning framework to adjust scRNA-seq data for ST analysis.
- To mitigate biases in cell-type decomposition caused by distribution discrepancies between scRNA-seq and ST data.
- To enhance the accuracy of cell-type proportion estimation in spatial transcriptomics.
Main Methods:
- Developed an instance-based transfer learning framework to adjust scRNA-seq data using ST data.
- Evaluated the impact of raw versus adjusted scRNA-seq data on cell-type decomposition using eight leading decomposition methods.
- Utilized both simulated and real-world spatial transcriptomics and scRNA-seq datasets for comprehensive validation.
Main Results:
- Data adjustment effectively reduced distribution differences between scRNA-seq and ST datasets.
- The adjusted scRNA-seq data significantly improved the performance of multiple cell-type decomposition methods.
- Enhanced decomposition accuracy led to a more precise mapping of cell type spatial organization.
Conclusions:
- Instance-based transfer learning is crucial for correcting distribution shifts in integrative scRNA-seq and ST analyses.
- Data adjustment prior to decomposition enhances the reliability and precision of spatial cell-type profiling.
- This approach provides valuable guidance for improving cell-type decomposition in spatial transcriptomics research.
