Accuracy improvement in protein complex prediction from protein interaction networks by refining cluster overlaps

Table 3 Clustering results of five density-based approaches and their accuracy on DIP data

method	number of clusters	average overlap rate	average f-score
CFinder	172	0.199	0.602
MCODE	272	0.000	0.456
DPClus	449	0.160	0.473
Entropy	1294	0.060	0.309
Modified-Entropy	110	0.099	0.485

We tested five density-based graph-clustering approaches on the yeast protein-protein interaction data from DIP. Their clustering results and the average overlap rates are shown. The overlap rate of a cluster indicates the proportion of overlaps in the cluster on average. As accuracy of each method on predicting protein complexes, we measured the average f-score of the clusters comparing to protein complexes experimentally determined.

ISSN: 1477-5956