Protein Network Topology Metric Conservation: From Yeast to Human

One of the most surprising results of the human genome project and the current sequencing efforts of other organisms is the remarkable similarity of the number of genes between species: the number of genes in C. elegans is of the same order of magnitude as that of Homo sapiens. It has been conjectured that these surprising results can be explained by a layer of protein-protein interactions, responsible for the expected difference in functional richness between worms and humans. For example, as the number of proteins n increases, the number of potential interactions increases proportional to n2. The quantity of interactions does appear to vary between organisms of different complexity. But, are there some constant parameters in these networks of interactions? As with the (Fibonacci) golden ratio ~1.618 (which appears in nature everywhere from flower pedals to shell patterns), perhaps there is a constant of nature with regard to biological network topology statistics. While it is not expected that such an absolutely invariant metric emerges, evolutionary pressures may have led to certain constraints on network parameters. By comparing yeast and human interaction data, we find that while the number of proteins and interactions are very different, topologic properties such as the average minimum distance between nodes, the power-law degree exponent, and the power-law proportionality constant appear to be relatively conserved.