Statistics on SOCs. (A) Number of species in the obtained SOCs. (B) Distribution of SOC sizes (number of clusters in SOC). (C) Phyletic patterns of SOCs. Each row represents an observed phyletic pattern, where white indicates presence and black indicates absence. The first seven columns to the left correspond to the different species. The eighth column depicts the frequency of the corresponding pattern, coded according to the color-bar on the right-most column. (D) Fractions of clusters participating in SOCs. Four bars are shown for each species (from left to right): fraction of clusters which are members of a SOC; fraction of clusters which are members of a SOC and pass the false negative filtering criterion (Supplementary Data); fraction of species-specific clusters; and fraction of species-specific clusters after filtering. (E) Phylogenetic trees relating the participating species. The larger tree (on the right) was reconstructed based on co-membership in SOCs. The tree on the left is based on conservation of single interactions. Percentages indicate the reproducibility of each branch in a bootstrap analysis (if different from 100%). Species names are abbreviated as follows: H. sapiens (H.s), D. melanogaster (D.m), C. elegans (C.e), S. cerevisiae (S.c), P. falciparum (P.f), E. coli (E.c) and H. pylori (H.p).

Overview of the SOC inference pipeline and subsequent analysis. (A) PPI networks of seven species are subject to cluster detection. (B) The obtained clusters are organized into orthologous groups based on sequence similarities of their member proteins. The subsequent analyses are done both on the cluster-level (frame C) and on the single protein-level (frame D). (C) The phyletic pattern of a given cluster is obtained according to the species present in the SOC it belongs to. Based on these patterns, we developed two novel measures for characterizing the evolution of clusters: PCL—an estimate for the rate at which the cluster was lost in evolution, and evolutionary age—the estimated emergence time of the cluster relative to the lineage split events in the phylogeny of the investigated species. The emergence time is estimated as the latest common ancestor (LCA) in the phylogenetic tree. (D) We study the correlation between the evolutionary properties of a cluster and those of its member proteins by comparing their evolutionary ages and loss rates. We also map the evolutionary cores of the SOCs to evaluate the role of gene duplication in their formation.

Case studies of clusters and SOCs. (A) The manually curated COPI in yeast. A cluster-level view of the corresponding SOC is given in (a.1). In this view, each node represent a whole cluster and edges represent putative orthology relations between clusters. The nodes are color coded according to their respective species (as in Figure 2). The yeast COPI complex is highlighted by an orange frame and its immediate neighborhood is highlighted by a purple line. A protein-level view of the highlighted subgraph in (a.1) is given in (a.2). In the protein-level view, the nodes represent single proteins, and edges represent either homologies (between species, red edges) or PPI (within species, blue edges). (B) A yeast-specific mating-type determination cluster. (C) A chromatin modification SOC, whose evolutionary core is enriched with duplicated, self-interacting proteins. A cluster-level view of the SOC is shown in (c.1). A protein-level view of the highlighted subgraph in (c.1) is given in (c.2). The proteins composing the putative core of the SOC are highlighted by a red frame. Evidently, the homology relations between the participating clusters are based on two conserved sets of duplicated proteins that substantially cover the putative core. The upper set is comprised of a set of paralogs from human (TBL1XR1, TBL1X, RBBP4) and single representatives from yeast and worm (TUP1 and K07A1.12, respectively). Similarly, the lower set is comprised of a set of paralogs from human (HDAC1/2/3/9) and single representatives from yeast and worm (RPD3 and C53A5.3, respectively).