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

BMC Bioinformatics

Table 7 The effect of similarity data.

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

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.

ISSN: 1471-2105