SPACEL: A Comprehensive Toolkit for Spatial Transcriptomics Data Analysis and Integration
Introduction
Spatial transcriptomics (ST) technologies have revolutionized the field of biology and medical research by enabling the detection of the spatial distribution of transcriptome in histological tissue slices. This groundbreaking technology allows researchers to probe specific transcripts or perform sequencing to gain insights into the transcription level in cells, predict cell types, and construct three-dimensional (3D) structures of tissues. However, the analysis of ST data can be challenging, especially when multiple slices need to be analyzed jointly using existing toolkits. Manual assembly of slices and construction of 3D structures can be time-consuming and prone to errors.
SPACEL: An Integrated Toolkit for ST Data Analysis
To address these challenges, a research team led by Prof. QU Kun from the University of Science and Technology (USTC) of Chinese Academy of Sciences (CAS) developed a novel computational tool called SPACEL (Spatial Architecture Characterization by Deep Learning). SPACEL is a comprehensive toolkit that integrates three modules, Spoint, Splane, and Scube, to automate the process of building 3D panoramas of tissues from ST data.
Spoint: Cell Type Deconvolution
Spoint is designed to perform cell type deconvolution, a crucial task in ST analysis. It predicts the spatial distribution of cell types based on the gene expression profiles measured in ST data. Spoint utilizes a combination of simulated pseudo-spots, neural network modeling, and statistical recovery of expression profiles to ensure the robustness and accuracy of its predictions.
Splane: Spatial Domain Identification
Splane employs a graph convolutional network approach and an adversarial learning algorithm to identify spatial domains by jointly analyzing multiple ST slices. It leverages the relationships between neighboring spots to capture the spatial context and identify distinct regions within the tissue.
Scube: 3D Tissue Reconstruction
Scube automatically aligns the ST slices and constructs a stacked 3D architecture of the tissue. It utilizes a deep learning-based approach to learn the transformation parameters between slices and integrates them into a coherent 3D representation.
Performance Evaluation
The researchers evaluated SPACEL’s performance on 11 ST datasets, totaling 156 slices, generated using various technologies such as 10X Visium, STARmap, MERFISH, Stereo-seq, and Spatial Transcriptomics. SPACEL demonstrated superior performance over other methods in three core analytical tasks: predicting cell type distribution, identifying spatial domains, and reconstructing three-dimensional tissue structures.
Conclusion
SPACEL provides a valuable integrated toolkit for ST data processing and analysis. Its ability to automate the construction of 3D tissue structures from raw data makes it a powerful tool for researchers employing ST technologies. SPACEL has the potential to accelerate discoveries in biology and medical research by enabling a deeper understanding of the spatial organization and interactions of cells within tissues.
Harness the power of SPACEL today and unlock the full potential of your spatial transcriptomics data!