Fig. 3: Differential expression analysis performed on CML state-space disease states.

A The three stable critical points of the CML state-space (\({c}_{1},{c}_{3},{c}_{5}\)) were used to group the CML samples into Early (Es), Transition (Ts), and Late (Ls) disease states based on their ___location in the CML state-space. The unstable critical points (\({c}_{2},{c}_{4}\); dashed lines) were used as the boundaries to define the three disease states. B Differential gene expression was performed both between the disease states and healthy control state (Hs) and between the disease states. The number of differentially expressed genes (DEGs) contained in the intersection between the comparisons was used to determine how similar the disease states were to each other. C Gene set enrichment analysis was performed on the log₂ fold change of the healthy vs disease states to determine which Hallmark gene sets were significantly enriched (between disease states shown in Fig. S3B). The results were summarized by showing a heatmap of the normalized enrichment score (NES) for all significant pathways (non-significant pathways are in grey) across the three disease states. The NES score shows the direction of expression change for each gene set. D Using the eigengenes (PC2 loading values) used to construct the CML states-space, we determined the contribution to CML of each gene that was a DEG in the disease state vs control comparisons. The CML contribution was determined by combining the eigengene value with the observed expression change for that comparison (right box). The total contribution for each DEG comparison was indicated by the mean vector (black arrow) of all DEGs in each comparison. E To visualize the contribution of each significantly enriched gene set, the total CML contribution (mean vector) for each comparison was plotted for each comparison (missing bars indicate non-significant gene set in that comparison).