Fig. 1

The Differential Multi-Information method. a Hypothetical scenario in which a putative Transcription Factor (TF) is activated by phosphorylation or de-phosphorylation through a Modulator (M). G1, G2 and G3 are three downstream targets of the TF. In presence of the Modulator (M) the downstream targets (G1, G2 and G3) become co-regulated through the active Transcription Factor (TF). b In absence of the Modulator (M) the downstream targets (G1, G2 and G3) are not co-regulated since the Transcription Factor (TF) is not active. c For each iteration of the DMI method a candidate modulator M is tested. First the GEPs are sorted according to the expression level of the modulator M and the GEPs subdivided in three (or more) subsets. The Differential Multi-Information (∆I) of the targets is computed always between the two subset where M expression is either “High” or “Low” by estimating the Renyi Multi-Information and taking its difference