Meet SPACEL: A New Deep-Learning-based Analysis Toolkit for Spatial Transcriptomics

Scientists traditionally examine tissues by analyzing the expression levels of genes in individual cells using a technique known as spatial transcriptomics (ST). Researchers gain insights into cells’ spatial organization and function by measuring the quantity of RNA in specific locations within a tissue. Spatial transcriptomics (ST) technologies have been instrumental in unraveling the mysteries of mRNA expression in individual cells while maintaining their spatial coordinates. However, challenges arise when multiple tissue slices need to be jointly analyzed, and the size of spots in ST slices hampers the resolution.

To overcome these limitations, a group of researchers headed by Prof. Qu Kun from the University of Science and Technology of the Chinese Academy of Sciences has created a solution called Spatial Architecture Characterization by Deep Learning (SPACEL). This toolkit has three modules—Spoint, Splane, and Scube—that combine to create a 3D panorama of tissues automatically.

The first module, Sprint, tackles the cell-type deconvolution task. It predicts the spatial distribution of cell types using a combination of simulated pseudo-spots, neural network modeling, and statistical recovery of expression profiles. This makes predictions accurate and powerful. The second module, Splane, utilizes a graph convolutional network (GCN) approach and an adversarial learning algorithm to identify special domains by jointly analyzing multiple ST slices. Splane uses adversarial training to remove batch effects over several slices and uses cell-type composition as input. Splane stands out for its innovative method of efficiently identifying spatial domains. The third module, Scube, automates the alignment of slices and constructs a stacked 3D architecture of the tissue. This is crucial in overcoming the challenges posed by the limitations of experimental ST techniques, allowing for a comprehensive understanding of the tissue’s three-dimensional structure.

The researchers applied SPACEL to 11 ST datasets totaling 156 slices and utilized technologies like 10X Visium, STARmap, MERFISH, Stereo-seq, and Spatial Transcriptomics. The researchers emphasize that SPACEL outperformed previous techniques in three fundamental analytical tasks—cell type distribution prediction, spatial domain identification, and three-dimensional tissue reconstruction.

Further, SPACEL demonstrated its superiority in cell type deconvolution, spatial domain identification, and 3D alignment against 19 cutting-edge techniques on simulated and real ST datasets, with its superior performance over previous techniques and simplified approach to accurately understanding ST data.

In conclusion, SPACEL’s introduction is a significant step in spatial transcriptomics. Its three modules provide researchers with a powerful tool to overcome the challenges associated with joint analysis of multiple ST slices, enabling precise cell type predictions, effective spatial domain identification, and accurate 3D tissue alignment. This tool allows for accurate 3D tissue alignment, cell type predictions, and efficient spatial domain identification.

Check out the Paper. All credit for this research goes to the researchers of this project. Also, don’t forget to join our 35k+ ML SubReddit, 41k+ Facebook Community, Discord Channel, LinkedIn Group, Twitter, and Email Newsletter, where we share the latest AI research news, cool AI projects, and more.