Skip to main content
Figure 1 | BMC Bioinformatics

Figure 1

From: Markov clustering versus affinity propagation for the partitioning of protein interaction graphs

Figure 1

Original unweighted protein interaction graph and graphs of curated complexes linked through their shared components. (a)Unweighted protein interaction graph comprising 1628 proteins (nodes) and 11 249 interactions (edges) generated from the 408 hand curated complexes of S. cerevisiae[28]. (b, c)two copies of a portion of the graph in (a), where complexes (nodes) are linked to one another whenever they share at least 2 components, and the node size is proportional to the number of unique proteins each complex contains. (b)and (c)have the AP and MCL clusters respectively mapped onto the curated complexes, so that pie charts show proportions of complex components that are annotated to the same AP or MCL cluster. The mapped clusters are computed from versions of the original unweighted network shown in (a) in which 20% of the edges were randomly added and 20% randomly removed. Complexes whose components distribute among many clusters appear as multi-colored pie graphs, whereas those that are annotated to the same cluster appear solid-colored. The bright red color indicates the proportion of components that were assigned to singleton clusters by the AP or MCL algorithm. All the comparisons were performed with partitions obtained by optimizing the MCL and AP parameters respectively (see Methods). The pie graphs were generated using the GenePro plugin[32] for Cytoscape[31].

Back to article page