Skip to main content
Fig. 1 | BMC Medical Genomics

Fig. 1

From: The similarity of inherited diseases (I): clinical similarity within the phenotypic series

Fig. 1

Assembly of a clinical similarity network. The D from OMIM (hexagons) are annotated with the DP from HPO (squares). As shown in the simplified hierarchical tree, the DP term 4 is the common ancestor of terms 2 and 3, whereas 0 is the root. Then, a typical similarity-searching algorithm retrieves all the possible Di − Dj pairs and the DP annotations that the two D share (dotted squares). The shared annotations can be either the identical DP that annotates both D or another term (often, the most informative common ancestor of two different DP). For instance, 1 is shared (by A and B) as identical term, while 4 is shared (by A and C) as common ancestor of 2 and 3. With these criteria, even two dissimilar D (e.g., B and C) will eventually share at least some (poorly informative) term (in this example, the root 0). At the end of the search, a non-weighted D-DP bipartite graph is assembled. Then, from the bipartite graph, a clinical similarity network is derived, by linking all the D that share DP annotations (the thickness of the edge being proportional to the similarity coefficients of the two linked D). Note that, for simplicity’s sake, the figure does not indicate the PS and that each pair of D is shown to share only one DP annotation

Back to article page