Output details

Graph spectral analysis of protein interaction network evolution

<28> Originality: The paper presents the first application of spectral graph theory to Approximate Bayesian Computation on networks.

Significance: The method allows for the use of Approximate Bayesian Computation on network models without resorting to summary statistics that may not be sufficient. Large problems are made tractable since there is no need to calculate computationally expensive network statistics. Applied to models of biological network growth, insights into evolutionary processes can be derived.

Rigour: The methodology is validated on simulated data and shown to reproduce known parameters, and applied to multiple real world biological data sets with similar results.