Fig. 3.

The CPI regime in Medusa. In the CPI regime, Medusa detects a relevant module of candidate objects based on a set of pivot objects, which belong to a different object type than the candidates. For example, given three diseases (pivots, black squares), we want to find potentially relevant genes (candidates, white circles), a task denoted with a dashed line. The figure shows six distinct semantics (solid lines) that exist in the fusion graph in Figure 1, which Medusa can choose or combine to identify the relevant module. For example, highlighted is an aspect that connects genes with diseases via chemicals (‘C’), pathways (‘P’), genes (‘G’), Gene Ontology terms (‘GO’) and exposure events (‘E’)