Skip to main content


Figure 2 | BMC Systems Biology

Figure 2

From: Supervised maximum-likelihood weighting of composite protein networks for complex prediction

Figure 2

Precision-recall graph for classification of co-complex edges using the five weighting schemes. (a) Classification of yeast co-complex edges. SWC and BOOST achieve the highest recall through data integration. TOPO has high precision for its top-scoring edges, but these are clustered in a few complexes. SWC achieves higher precision than STR, except when too many edges are considered. BOOST classifies edges categorically, giving high scores to one set of edges with about 50% recall and 35% precision, and low scores to the remainder. (b) Classification of human co-complex edges. Recall and precision for human is much lower than for yeast. TOPO has higher precision than SWC, but its predicted edges are clustered in fewer complexes. BOOST classifies edges categorically, and its high-scoring edges achieve 7% recall, with comparable precision with SWC. NOWEI has slightly higher precision than STR, which has the lowest precision.

Back to article page