Fig. 3

Information Theory-Based Analysis Suggests Majority of Genes Have Minimal Linear and Non-linear Relationships with Age Groups. a Kullback-Leibler (KL) divergence representation of SHH (top) and IL6R (bottom) distributions for adult and pediatric samples. b KL-divergence versus Log Fold Change plot, with previously identified therapeutic targets overlaid and all points colored by IHW-adjusted p-value, demonstrating relationship between calculated LFC and informational differences. Higher KL-Divergence values indicate that a gene has informational difference between age groups. As KL-Divergence is an asymmetric method, the scenario of having a pediatric prior (left) and an adult prior (right) are both shown. c Pearson Correlation Coefficient (PCC) vs Maximal Information Criterion (MIC) plot with previously identified targets overlaid. MIC scores the strength of a relationship from 0 (no relationship) to 1 (noise-free relationship) for genes between the age groups. Points A and B on the graph represent where genes should fall if they have a strong direct or inverse linear relationship with age groups. Values that have low PCC and high MIC scores indicate genes with non-linear (i.e. dynamic; one-vs-many) relationships. d Summary statistics of where genes lie on the PCC vs MIC plot along with their respective differential expression values as visualized in (b)