Fig. 2

Cluster analysis suggests that transcriptional signatures correspond with the underlying tissue but are a conglomerate of cells and cell types per spot. A–H Unbiased k-means clustering spatially allocated to their original capture area and corresponding unbiased uniform manifold approximation and projection (UMAP) clustering of the exact same clusters. CT, connective tissue. I Illustration of how unbiased clustering corresponded to the underlying histology. The green, pink, brown, and purple clusters appeared to be derived from connective tissue- and adipose-rich areas, respectively, while orange and blue clusters represented muscle fiber-rich areas. J Transcripts were typically derived from multiple cells (and cell types) per spot: K Cluster-specific myofiber and nuclei numbers per capture spot differed depending on the underlying tissue type. L As a practical example of manual tissue identification by transcriptional markers, the connective tissue marker COL1A1 was detected in most capture areas, but M connective tissue-rich areas could be highlighted by setting a marker expression threshold. Data in (K) are means and standard deviation. *p < 0.05; **p < 0.01; ***p < 0.001