Skip to main content

Table 7 The effect of similarity data.

From: Hubs of knowledge: using the functional link structure in Biozon to mine for biologically significant entities

Query Term Query Type Prominence Model Focused Subgraph Average No. Neighbors Average Consistent Neighbors Average Ratio Q(R) UROC (R)
autoimmune protein Hubs & Authorities With Sim 1413.76 1.98 0 439 11773
    No Sim 69.8 4.64 0.13 2062 57865
stromelysin protein Hubs & Authorities With Sim 338.52 27.98 0.083 10348 263293
    No Sim 214.91 19.3 0.22 6690 181282
  1. These quality results were calculated for the stromelysin and autoimmune focused subgraphs when searching for proteins. The Hubs & Authorities values were computed using the Max scoring method. To compare these results to our previous results we recomputed all performance measures for the focused subgraphs that include similarity relations, but using the ranking that was produced without considering these relations. Interestingly, when using similarity data, the top scoring entity for the 'stromelysin' query is a protein (docid 986092) that does not contain the query term in its definition, nor do the DNA sequences, the UniGene clusters, and the enzyme family that are related to this protein. However, this protein, membrane type 5 matrix metalloproteinase, is significantly similar to many stromelysin proteins.