Figure 5
Reproducibility of the predictions in similar data sets. For every CLCD prediction (A → S → T), all possible networks were scored in an independent data set, and prediction is consistent if the same network is selected by the scoring scheme. In contrast, if the highest scoring network contradicts the predictions (i.e. includes a S ← T edge), the prediction is considered conflicted. Predicted networks are found consistent in a variety of data sets, measuring the same variables and activators in (right) 8 distinct time points, (middle) 8 different healthy donors and (right) bone marrow data from two healthy donors in an independent study24. Asterisks denote statistical significance at 0.05(*) or 0.001(**). A one-tailed t-test is used to compare consistent predictions to consistency at random.
