Figure 1.
An imaginary network with artificial experimental data values is shown (e.g. relative gene expression values) on the left. Node A was assigned a value of 5, nodes G, H, I, J, K and L were assigned 2, and all the other nodes were assigned 1. A transition probability matrix P was constructed using the input data values and the network, with transition probabilities between adjacent nodes reflecting their data values (colors in the matrix reflect transition probabilities P(i→j) according to the color key). Final visitation and flux values reflect the level of coherence between the experimental data of genes and their relative positioning within the network. Note that node colorings in the network on the right reflect relative visitation probabilities of nodes, and line colors of edges reflect the flux values according to the same color scale.
Figure 2.
NetWalk analysis of low and high-dose doxorubicin response in MCF7 cells.
A) Apoptosis levels in MCF7 cells after 24 hours of stimulation with indicated doses of doxorubicin as measured by FACS analysis of DNA content (see Methods). B) FACS analysis of viable cells as indicated by loss of Rhodamine 123 staining(see Methods). C–D) Plots of interactions with lowest(B) and highest (C) EF values in samples treated with 1 µM doxorubicin for 24 hours relative to control. Nodes are colored according to their gene expression change relative to control according to the color key. Edge coloring reflects type of interaction, PPI: protein-protein interaction, TF-target: gene regulation, FS: functional similarity. The distribution plot of all EF values is shows at the bottom.
Figure 3.
Comparison of coherence of node values in highest scoring networks.
A) Boxplots of gene expression change values (1 µM DOX, 24 hours relative to control) of nodes in networks generated by different cutoffs of EF values, or in networks generated by Ingenuity Pathway Analysis software using different gene expression value cutoffs for the focus gene set (see Methods). B) Heatmaps showing position of genes in the networks in A in the whole data distribution. Positions of genes in the respective networks are indicated by a white line. C) A network of nodes generated by Ingenuity Pathway Analysis software with focus gene set using 1.5 as cutoff. Since original network plots in IPA lack node colorings for intermediate genes (non-focus genes), we extracted all nodes in the IPA-generated network and re-plotted them using our network, where we colored all nodes by their gene expression change.
Figure 4.
Clustering analysis of EF values in each condition.
A) Heatmap of highest and lowest EF values in each condition. Clustering was done using Ward's method in R. B–C) Networks corresponding to K3 (B) and K4 (C). Node colorings are according to 24h of 1 and 10 µM DOX treatments, respectively. Edge colorings are as in Figure 2C.
Figure 5.
p53-target cell cycle regulatory genes are specifically repressed during apoptosis.
A–B) Network plot of interactions in K3 (see Figure 4) related to cell cycle regulation. Nodes colored according to gene expression changes at 10 (A) or 1 µM (B) doxorubicin treatment. C) Western blots of p53, p21 (CDKN1A gene product) protein levels over a time course after 1 and 10µM doxorubicin treatment. Actin levels shown as control.