Skip to main content
Fig. 1 | BMC Systems Biology

Fig. 1

From: Annotating activation/inhibition relationships to protein-protein interactions using gene ontology relations

Fig. 1

Method overview. a Input was an undirected, unsigned PPI network. b, c, d For each undirected, unsigned PPI, direction was predicted. We trained two logistic regression models that predicted whether the signal can flow in left-to-right direction and right-to-left direction, respectively. Feature vector was composed of pairs of GO terms between which regulates, positively regulates, or negatively regulates relation holds. e A directed, unsigned PPI was obtained as a result of direction prediction. f A sign of PPI was predicted for each directed PPI. We trained two logistic regression models that predicted whether a directed PPI can act as activation or inhibition, respectively. Feature vector was composed of a pair of GO terms between which positively regulates or negatively regulates relation holds. g, h As a result of direction prediction and sign prediction, we annotated effect types to PPI network. Abbreviations: GO, gene ontology; PPI, protein-protein interaction; LR, logistic regression

Back to article page