Proteins are one of the most fundamental building blocks of life. These macromolecules are important for many biological processes such as molecule transport, DNA replication, enzyme catalysis etc. Protein contact networks are a network-based representation of the three-dimensional structure of proteins.

In this work I highlight the peculiar topologic properties of Protein Contact Networks by focusing on the spectral characteristics of their normalized laplacian matrix. I then discuss the problem of generating new networks that are, from a spectral perspective, as similar as possible to a set of experimental Protein Contact Networks. The proposed generation mechanism is a two-step process: first, a simple network generative model is defined as a baseline; in the second step the generated networks are iteratively optimized by means of an evolutionary optimization algorithm, in order to improve their similarity to the target set of experimental networks.