Comparison of model predictions with published experimental evidence.

Each row compares a model prediction about a genetic interconnectivity parameter Tij, representing the regulation of gene i by gene j, with published experimental evidence. Comparisons of the same parameter to multiple papers are listed in separate rows. Tij is listed as T_Genei_Genej. The prediction column lists the type of regulation inferred by the model. It shows activation or repression when the first quartile of the distribution of the inferred parameter is positive or if the third quartile of the distribution is negative respectively (Fig 4). The prediction column shows “sign not constrained” when the interquartile range spans negative and positive values. The experiment column lists that type of interaction established in the paper. If the paper describes evidence only of binding but not whether the target is activated or repressed, then the entry is “binding”. Negative experimental results are listed as “no effect found”. The entry in the experiment column is “not found” if we were not able to find any published tests of the parameter in question. The “Status of prediction” columns lists whether the evidence matches the prediction or not. “confirmed” implies agreements, while “incorrect prediction” implies disagreement. An asterisk indicates that conflicting experimental evidence was found. Conflicting evidence was found for the regulation of Gata1, Spi1, and Gata2 by Gfi1 (Moignard et al. [53]). Situations where no evidence was found or the paper reported negative results are listed as “undetermined”. The “Type of evidence” column classifies the evidence as genetic, protein-protein interaction, cis regulation, or functional cis regulation. Genetic evidence involves genetic manipulation of the predicted regulator followed by a characterization of the target’s expression and usually cannot distinguish between direct and indirect effects. Protein-protein interaction implies biochemical evidence of direct protein-protein interactions. cis regulatory evidence indicates direct interactions by identifying regulatory elements or binding sites potentially bound by the predicted regulator but does not establish a functional relationship between binding and the expression of the target gene. Functional cis regulation goes a step further and manipulates the binding sites and measures reporter or target expression to provide evidence that the binding of the regulator has functional impacts. The “Organism/Cells” and “Citation” columns list the organism or cells in which the interaction was tested and the DOI URL for the paper respectively. (TSV)