Reverse Causal Reasoning

The RCR algorithm was the implemented back in 2012 by the BEL group in Selventa. Similarly to Heat diffusion, a graph and mappeable are required as input. However, as opposed to Heat diffusion, the rationale behind it is to test the concordance of the data to what the knowledge graph would predict. As an example, statistics about whether a node regulates a set of downstream nodes or not by inferring statistics from the concordances.

As a demonstration, we can show how the upstream nodes regulate a series of nodes with a predicted effect. After overlapping the data in each of the nodes, the table on the right side of the figure shows whether there is a match or not between the prediction and what has been measured in an experiment.

Figure: RCR algorithm