Table 3. Examples of highly ranked nodes for some of the cell types used for learning of the NN.
| Cell type | TF/PPI/dense node | Corrected P-value | GO function/reference |
|---|---|---|---|
| Stem cells (ES) | Dense 35 | 3.25E-07 | Cell differentiation |
| Stem cells (ES) | Dense 24 | 4.44E-15 | Factor: E2F-3; (45) |
| Stem cells (ES) | Dense 24 | 5.78E-12 | Factor: IRF6; (46) |
| Stem cells (ES) | Dense 24 | 2.53E-08 | System development |
| BMDC | Dense 10 | 1.56E-05 | Immune system process |
| BMDC | Dense 67 | 1.59E-11 | Positive regulation of immune system process |
| BMDC | Dense 36 | 2.06E-06 | Response to cytokine |
| Fibroblast | ppi 223 | 3.16E-06 | Focal adhesion (47) |
| Fibroblast | ppi 301 | 2.85E-20 | Acetyltransferase complex (48) |
| Zygote | ppi 10 | 6.85E-07 | Regulation of cell proliferation |
| Zygote | ppi 280 | 7.36E-09 | Cell junction |
| Zygote | TF: foxd3 | X | TF: foxd3 (49) |
Some of the nodes were based on TF–gene interactions and thus represent a specific TF (Foxd3). For these, we rely on the function of the TF to characterize the node in the table. Other nodes are either based on PPI (e.g. PPI223) or on groupings learned by the algorithm (e.g. dense24). For thes, we performed GO analysis on the set of highly ranked genes for these nodes. Several other relevant TFs and functions were found for other cell types as well. See Supporting Website for complete list.