HeartMAP: A multi-chamber spatial framework for cardiac cell-cell communication

Figures

• 

  Fig. 1. demonstrates the proportional distribution of cells across all four cardiac chambers in the dataset, showing clear chamber-specific patterns that remained consistent across different analytical scales. When the dataset was scaled to 50,000 cells for computational efficiency, the proportional relationships were maintained with remarkable consistency: RA contained 14,192 cells (28.4 %), LV had 13,510 cells (27 %), LA contained 13,186 cells (26.4 %), and RV contained 9112 cells (18.2 %). This consistency validates the robustness of the chamber distribution patterns and confirms that the observed differences reflect genuine biological variation rather than sampling artefacts.

• 

  Fig. 2. (a) presents a pie chart showing the relative abundance of major cardiac cell types identified in the 10,000-cell analysis subset, with each coloured segment representing a distinct cellular population. The distribution demonstrates successful identification of all major cardiac cell types, with cluster 3 representing the largest population at 25.5 % (2548 cells), followed by cluster 6 at 19.8 % (1979 cells), and cluster 7 at 11.9 % (1194 cells). The remaining clusters (0, 1, 2, 4, 5, 8, and 9) show progressively smaller populations, with cluster 5 representing the smallest identified population at 1.2 % (120 cells). This exhaustive cellular diversity confirms robust representation of the major cardiac cell populations, including cardiomyocytes, fibroblasts, endothelial cells, smooth muscle cells, and immune cells that characterise human cardiac tissue architecture. (b) shows the distribution of total UMI (Unique Molecular Identifier) counts per cell across the dataset, representing the overall transcriptional activity captured in each cell. The distribution shows a characteristic right-skewed pattern with most cells containing between 717 and 2673 UMI counts, indicating robust…

• 

  Fig. 3. Heatmap of chamber-specific marker gene expression across 287,269 cells (log2(CPM+1), Wilcoxon p < 0.05, FDR < 0.05, |log2FC= > 1). The visualisation demonstrates how atrial chambers share common endocrine functions through NPPA and MYL7 expression while maintaining chamber-specific adaptations, and how ventricular chambers exhibit elevated expression of contractile proteins like MYH7 and FHL2 that support their specialised high-pressure pumping machinery.

• 

  Figure 4. displays the correlation matrix visualisation that reveals these hierarchical relationships between chambers, with colour-coded representation demonstrating how functional similarity translates into transcriptional correlation patterns across the cardiac architecture.

• 

  Fig. 5. UMAP visualisation of communication hub scores across cardiac cells.

• 

  Fig. 6. Cell-type communication specificity.

• 

  Figure 7. presents the quantitative analysis of differentially expressed genes for each chamber pair comparison, demonstrating the extensive molecular diversity that underlies cardiac chamber specialisation and the wealth of potential therapeutic targets identified through this analysis.