< Back to Article

Connectivity Homology Enables Inter-Species Network Models of Synthetic Lethality

Fig 1

A) An illustration of connectivity homology (CH). Each node is described by two parameters (degree [deg.] and betweenness centrality [bet.cent.]) at three levels: low, medium, and high.

Certain nodes have the same vectors (Node B/Node 2/Node 3); these nodes can be said to be connectivity homologous (CH). Other nodes do not (Node A/Node 1); these are non-connectivity homologous (non-CH). B) Schematic of the SINaTRA algorithm. We begin with the PPI networks of both our source and target species, calculate the network parameters (independently), and normalize the values of all parameters. Next, we use machine learning methods on the normalized network parameters of our source species as well as experimentally derived labels of synthetic lethality to construct a species-independent model of SL. Finally, we apply this model to the normalized network data of our target species in order to attain SL predictions in our target.

Fig 1