Figure 1.
Flowchart of the proposed methods to estimate network robustness and response ability.
This flowchart delineates the process used to construct the gene regulatory network (GRN) and protein-protein interaction network (PPIN), and the subsequent estimation of network robustness and response ability by the NRM and RAM methods, respectively. The flow chart on the left represents the Case 1 study analysis of the GRN; the flow chart on the right represents the Case 2 study, where methods are applied to a PPIN. [AIC-Akaike Information Criterion].
Figure 2.
Multiple regulatory loops of a GRN associated with aging-related pathophysiological phenotypes.
This network includes the following sixteen genes: FOXOs, NF-κB, p53, SIRT1, HIC1, Mdm2, Arf1, PTEN, PI3K, Akt, JNK, IKKs, IκB, BTG3, E2F1, and ATM. Blue arrows indicate activation; blunt red arrows indicate suppression.
Table 1.
Ordinary differential equations of gene regulatory networks in Figure 2 for sixteen genes associated with aging-related pathophysiological phenotypes.
Figure 3.
The locations of the sixteen eigenvalues of different tissues at the young and aged stages, respectively.
Some eigenvalues of the interactive matrix A for the thymus (A) and spinal cord (B) are located together at similar regions near the unit circle |Z| = 1.
Table 2.
Estimated parameters of gene regulatory networks in Table 1 with sixteen genes in the thymus (A) and spinal cord (B) at the young and aged stages.
Table 3.
Estimated network robustness (ηo) and response ability (δo).
Figure 4.
Refined protein-protein interaction network.
The figure shows the refined PPIN with effective protein interactions under oxidative stress in fibroblast (A) and HeLa cells (B). A dynamic PPI model and model selection method, Akaike Information Criterion (AIC), are used together to prune the rough PPIN using the time series microarray data to delete unrealistic and false positive PPIs in the PPIN.