Figure 1.
Alternative three-way network topologies including a Transcription Factor (TF), a Target gene (Tg) and a Modulator gene (M).
(A) depicts the independent regulation of the target gene by a modulator and a TF; (B) describes a three-way interaction between the TF, the target gene and the modulator.
Figure 2.
Schematic representations of the CINDy algorithm.
A collection of gene expression profiles is required to calculate Conditional Mutual Information between lists of modulators, transcription factors and putative target genes, with the final output of inferred modulation events.
Table 1.
Default parameters used for running MINDy.
Figure 3.
Comparative performance of MINDy and CINDy.
Precision and recall values are compared in the B-cell lymphoma dataset (panel A) and Lung Adenocarcinoma dataset (panel B), calculated by matching the predictions with a gold standard dataset set obtained from four different databases of experimentally validated PPIs between modulators and transcription factors. Precision and recall are further compared at different robustness threshold for MINDy (blue line) and CINDy (red line) in the B-cell dataset (panel C) and in the Lung dataset (panel D, see Materials and Methods).
Figure 4.
Example of novel prediction by CINDy.
Proposed mechanism for modulation of HMGA1 by CDK2.