Bio-molecular networks became popular during the last decade with the sudden increase in data about protein and other biomolecule interactions. Network representations of these data has allowed graph theory to be applied to bio-molecules and model the pairwise relationships between them where proteins and enzymes are characterized as network nodes and the interactions between them as network links (edges). The two types of networks typically constructed are protein-protein interaction maps and metabolic networks.
Metabolism is a fundamental process in all organisms that is carefully controlled to ensure their survival, but enzymes and proteins in metabolic pathways undergo ...view middle of the document...
Betweenness is the frequency with which a node is located on the shortest path between all other nodes. Metabolic flux is the rate at which reactants are converted into products by chemical reactions.
Evolution of a network i.e., addition and deletion of nodes and/or edges occur due to genetic changes. Gene duplication entails the addition of a network node and its associated edges whereas gene loss indicates that the node and its associated edges are lost. Mutations that disturb gene regulation, point mutations, deletions, insertions, inversions, shuffling, alternative splicing and domain accretion can also cause gain and loss of nodes and edges. On the other hand, network function may limit the kinds of mutations that can be accepted, and thus how genes evolve. It should be noted that the evolution of links is a lot more fine-tuned than that of nodes as links can change over time irrespective of whether or not nodes change.
DATASET AND METHODS
Greenberg et al studied adaptation and evolutionary constraint in twelve strains of Drosophila that are closely related to (and include) Drosophila melanogaster. The maximum likelihood estimates of rates of non-synonymous amino acid substitutions dN, synonymous substitutions dS and their ratio, ω (ω = dN/dS) were used for each gene. Then positive selection tests were performed using PAML (Model M8 against M7). Fuzzy reciprocal BLAST of every D.melanogaster gene against all other genomes was used to estimate rates of gene duplication. The metabolic network and pathway assignment of D.melanogaster was obtained from KEGG and gene information & mutation phenotypes from FlyBase. Evolutionary rates were calculated for 447 genes. Partial correlations were used to assess the weight of each network parameter. Permutation tests were used to evaluate P values and for enzymes that belonged to multiple pathway groups and enzymes encoded by multiple genes, modified permutation tests were used.
Vitkup et al investigated how the topology of a metabolic network and its reaction fluxes affect the evolution of network genes through point mutations and gene duplications in the yeast Saccharomyces cerevisiae and three other yeast species. To elicit only genuine functional relationships, 14 most connected metabolites and co-factors were excluded from most calculations. This resulted in about 5% of network enzymes becoming disconnected from the network. Unlike the earlier study, this study used Ks, Ka and Ka/Ks values in their calculations. The calculations were based on the maximum likelihood models of Muse & Gaut and Goldman & Yang for estimating gene duplication rates, previously published studies and the maximum likelihood model of Yang & Nielsen for Ks, Ka and Ka/Ks. They also performed partial correlation analysis to relate network parameters.
Montanucci et al analyzed the evolutionary constraints of the N-glycan pathway by estimating the dynamics, dispensability of the pathway components, biological & topological...