Figure 4. Gene expression and Pearson correlation pattern analyses.

A.-B. PCA analyses with Immune DOWN and Cycle UP genes recapitulating the trajectory of human lung development. The lung developmental samples clustered tightly within each developmental stage, but not between different stages, implying that the differentiation of these genes might promote human lung development. C.-E. Heatmaps of Pearson correlation patterns for Immune DOWN and Cycle UP genes in C. the development stage, D. the precancerous progression stage and E. the cancer stage. Red rows represent Cycle UP genes, while blue rows represent Immune DOWN genes. The heatmaps indicated that the co-expression pattern was continuously disrupted from development to carcinogenesis. F. Density curves of Pearson correlation in three stages. A bimodal distribution was seen for the developmental stage, in contrast to the unimodal distributions of the progression and cancer stages. Furthermore, the cancer stage had a higher maximum density at a Pearson correlation of zero compared with the progression or developmental stages, also indicating deteriorating gene-to-gene co-expression.