Figure 1
From: Collective influencers in protein interaction networks
Topological and functional characteristics of collective influencers (CI) in human protein interaction networks. (a) In the table we present statistics of human protein interaction networks and their corresponding CIs. In particular, we accounted for binary and co-complex interactions as well as a combined interaction data set. Notably, the frequency distribution of the degrees of (non-)CI proteins in the combined network indicated that CI proteins were involved in a higher number of interactions. Furthermore, we observed that CI proteins were also frequently low-degree proteins. (b) Randomizing the set of CIs, we determined the enrichment of interactions between (non-)CI proteins. We found that interactions between CI proteins were enriched and appeared depleted between non-CI proteins in all networks. (c) CI proteins in the combined network composed a giant connected component with 4,639 CI proteins (dashed line). Such a result was significant when we randomized sets of CI proteins and determined the distribution of the sizes of the giant components thus obtained (P < 10−6). In (d), we randomized essential human proteins and observed that CIs were strongly enriched with essential genes in all networks. (e) Considering interactions between essential genes, we found that such interactions preferably appeared between CI proteins. (f) While CIs were strongly enriched with kinases, we found no significant enrichment of CIs with transcription factors in the combined network (inset). Randomizing sets of CI proteins in the combined network, we found that links between transcription factors and their corresponding targets were enriched when transcription factors were CIs. We obtained a similar result using kinase-substrate interactions.
