Information Content-Based Gene Ontology Semantic Similarity Approaches: Toward a Unified Framework Theory

Table 3

Comparison of performance of different approaches for GO MF ontology. This comparison is done using Pearson’s correlation with enzyme commission (EC), Pfam and sequence similarity, and resolution. Results are obtained from the CESSM online tool. The best scores among each group are in bold, and Nmax, Nunif, and Nunivers are suffixes indicating different IC normalization strategies, namely, the highest IC value, uniform, and GO-universal strategies, respectively.

Family	Approach	Similarity measure correlation			Resolution
		EC	PFAM	Seq Sim
Annotation	Resnik-Nmax	0.64381	0.49101	0.59663	0.55309
	Resnik-Nunif	0.64381	0.49101	0.59662	0.28872
	Nunivers	0.70697	0.47693	0.40945	0.41671
	Lin	0.67404	0.42844	0.36060	0.36583
	Li et al.	0.70287	0.46309	0.38823	0.44311
	Relevance	0.67618	0.42112	0.35081	0.39798
	GraSM-Lin	0.68125	0.44009	0.37243	0.38321
	GraSM-Nmax	0.65180	0.49844	0.60405	0.37213
	GraSM-Nunif	0.65180	0.49844	0.60405	0.28859
	GraSM-Nunivers	0.71257	0.48638	0.41889	0.43191
	XGraSM-Lin	0.70480	0.53732	0.47682	0.43007
	XGraSM-Nmax	0.67136	0.58792	0.70911	0.36781
	XGraSM-Nunif	0.67136	0.58792	0.70911	0.28524
	XGraSM-Nunivers	0.71965	0.55251	0.48988	0.47064
Topology	Wang et al.	0.64327	0.46102	0.37272	0.34873
	Zhang et al.	0.68296	0.43453	0.35581	0.38646
	GO-universal	0.67661	0.47000	0.38190	0.43772