Here, we used the predicted A. thaliana and P. trichocarpa 0.65 confidence PPI networks to conduct a preliminary analysis of the relationship between WGD duplicate retention and 2 properties of the predicted network: the fraction of genes to which a given gene is connected (degree centrality) and the fraction of neighbors retained following the same WGD (duplicated neighbors).
In the model realization with reinforcement the probability of discovering an edge is proportional to the sum of the degrees of the proteins it connects such that edges connecting higher degree proteins are more likely to be discovered as indicated by the weight of the edge line.
We argue that this straightforward and standard method may lead to terrible bias, since in real biological and information networks, missing links are more likely to be links connecting low-degree nodes.
In addition, it is also observed that (i) edges connecting high-degree nodes are more important, as they connect many nodes and may be relevant to the global structure property of the network (H., 2006); and (ii) a pair of two nodes with a large number of common neighbors are more likely to be related (Liu, 2009).
Another important property revealed by the CPD is disassortative mixing, i.e., hubs (highly connected nodes) tend to connect low degree partners [12], [13], which results in the continuous bright ring distributed evenly around the whole circumference.
The last column lists highly connected genes (degree ≥35) of the corresponding module ordered alphabetically.
Therefore, topological parameters such as the clustering coefficient, connected components, degree distribution and average shortest path were analyzed by Network Analyzer in Cytoscape software.
Networks were overall highly connected, with degree distributions left-skewed and most genes having >400 neighbors (Additional file 1: Figure S1).
In 2003 and 2012, the tendency is to have high degree nodes connected to low degree nodes (descending curves).
