Spatial Patterning Analysis of Cellular Ensembles (SPACE) Detects Complex Spatial Organization at the Cell and Tissue Levels
Spatial Patterning Analysis of Cellular Ensembles (SPACE): Revealing Complex Spatial Patterns in Biological Tissues
Academic Background
The function of biological tissues depends on the interactions of cells, molecules, and structures in three-dimensional space. Understanding these spatial organization patterns can provide new insights into pathologic and physiologic processes. However, existing spatial analysis technologies mostly detect only simple patterns and cannot capture complex spatial relationships. To address this challenge, researchers have developed an open-source analysis tool called “Spatial Patterning Analysis of Cellular Ensembles” (SPACE), which aims to comprehensively detect and describe complex spatial patterns in biological tissues using information-theoretic calculations.
The development of SPACE was driven by the limitations of existing analytical tools. Many spatial analysis tools can only detect single or paired biological elements but are unable to handle complex patterns involving three or more elements. Additionally, discrete microenvironment (MEs) analysis methods cannot capture continuous features such as gradients or orientations. The introduction of SPACE fills this gap by detecting spatial patterns of any complexity and providing detailed descriptions of these patterns.
Paper Source
This paper was written by Edward C. Schrom et al., with primary authors from the National Institute of Allergy and Infectious Diseases (NIAID) and other related institutions. It was published in the Proceedings of the National Academy of Sciences (PNAS) on February 4, 2025. The research was supported by the National Institutes of Health (NIH) and other organizations.
Research Methods and Workflow
1. Data Input and Preprocessing
SPACE supports various data input formats, including molecular expression images, segmented cell images, pixel-level structural data, and tables of cellular centroid coordinates. These data can be two-dimensional (2D) or three-dimensional (3D). SPACE collects information on biological elements within each neighborhood by drawing circular or spherical neighborhoods across different data layers and generating statistical censuses. To control for the effects of tissue geometry and compositional abundances, SPACE also generates randomized censuses as controls.
2. Detection and Quantification of Spatial Patterns
SPACE uses mutual information (Mutual Information) from information theory to quantify deviations from random spatial patterning. For single specimens, SPACE calculates “cis mutual information” (cismi) to measure the difference between observed spatial patterns and random distributions within a specimen. For multiple specimens, SPACE calculates “trans mutual information” (transmi) to compare spatial patterns across sample groups. By exhaustively exploring all possible combinations of biological elements, SPACE identifies and ranks the most significant spatial patterns.
3. Characterization of Pattern Features
SPACE employs self-organizing map (SOM) techniques to find curves of highest neighborhood density in multidimensional space and projects them into lower dimensions, generating covariation plots. These plots show continuous variation features of biological elements, such as gradients, orientations, and context-dependent associations. Researchers can manually gate microenvironments and further analyze the unique spatial patterns within these regions.
4. Data Validation and Result Analysis
SPACE has been validated on multiple datasets, including mouse lymph nodes, mouse tumors, human intestinal cancer spatial transcriptomics data, and human tuberculosis granulomas. Through these analyses, SPACE not only successfully identified known spatial patterns but also discovered new complex patterns. For instance, in mouse lymph nodes, SPACE revealed opposing gradients of CD4 and CD8 T cells in the paracortex; in human intestinal cancer, it found complex co-occurrence patterns of IG4 and SFRP2 transcripts; in tuberculosis granulomas, it identified associations between CD68 macrophages and immunoregulatory molecule expression.
Key Research Findings
- Mouse Lymph Node Analysis: SPACE successfully identified the distribution of B cells in the dark zone (DZ) and light zone (LZ) of germinal centers (GC) and revealed the opposing gradient of CD4 and CD8 T cells in the paracortex. These patterns showcase the complex spatial organization of immune cells in lymph nodes.
- 3D Mouse Tumor Analysis: SPACE analyzed the spatial distribution of CD4 T cells, CD8 T cells, and blood vessels, finding that CD4 T cells tend to locate closer to blood vessels, consistent with literature on tumor growth and immune regulation.
- Human Intestinal Cancer Spatial Transcriptomics Analysis: SPACE identified co-occurrence patterns of multiple transcripts, including MGAM2, GPX2, and PPP1R1B. Additionally, it uncovered complex relationships between IG4 and SFRP2 transcripts, relevant to cancer immunity.
- Human Tuberculosis Granuloma Analysis: SPACE analyzed tuberculosis granulomas under different clinical statuses, finding that the aggregation pattern of CD68 macrophages correlates with disease severity and revealing local changes in immunoregulatory molecule expression.
Research Conclusions
SPACE is a versatile spatial analysis tool capable of detecting spatial patterns of any complexity in biological tissues. It is compatible with various data formats and experimental methods and can directly compare spatial patterns across different sample groups. Through SPACE, researchers can extract key biological tissue patterns from high-dimensional spatial data, providing new insights into immunology, cancer research, and infectious diseases.
Research Highlights
- Comprehensiveness: SPACE can detect and describe complex spatial patterns involving any number of biological elements, filling gaps left by existing analytical tools.
- Versatility: SPACE supports multiple data formats and experimental methods, applicable to both 2D and 3D data, and can integrate multiple data sources.
- Innovation: Using information theory and self-organizing maps, SPACE achieves systematic detection and detailed description of complex spatial patterns for the first time.
- Application Value: SPACE offers powerful analytical tools for immunology, cancer research, and infectious diseases, aiding in the discovery of new mechanisms and potential therapeutic targets.
Additional Valuable Information
All data and analysis code from this study are publicly available for use and modification by other researchers. Additionally, the research team provides detailed tutorials and example data to help users quickly master the use of SPACE.
Through SPACE, researchers can extract key biological tissue patterns from complex biological data, providing new tools and methods for understanding tissue function and pathological processes.