Fig. 4

Temporal ordering of differentiating primary human myoblasts using transcript-compatibility counts. a A minimum spanning tree (MST) was drawn through the 271 cells using Jensen-Shannon distances computed between 1,101,805-dimensional vectors of TCCs of cells in the dataset. Following the longest path does not show a clear cell differentiation pattern. b Affinity propagation clustering generated seven clusters (after collapsing spurious clusters with less than five cells into their nearest neighboring cluster), and an MST was drawn through the centroids of the clusters. Using the labels from Trapnell et al. [12], the longest path shows a differentiation pattern from proliferating cell (red) to differentiating myoblast (blue). The MST also shows how some proliferating cells alternatively differentiate into interstitial mesenchymal cells (green). c The cells were then clustered into three groups based on their transcript-compatibility counts, and the MST from b was relabeled using these new cell types. d The expression levels of the genes MYOG, CDK1, and PDGFRA were analyzed for the three TCC clusters. MYOG, CDK1, and PDGFRA show greater expression for centroids from clusters with greater proportions of TCC cell types 1, 2, and 3, respectively. For each gene, a histogram over each centroid shows how expression level evolves with the differentiation process. CDK1, MYOG, and PDGFRA being markers for proliferating cells, differentiating myoblasts, and interstitial mesenchymal cells indicates that the clustering and centroid-ordering based on TCC captures intermediate steps of the human myoblast differentiation trajectory