Simple Rules Imposed on a Primitive Cubic Lattice Robustly Generate Structures that Mimic Features of Real Proteins

We introduce a set of simple and well-defined rules, which produce protein-like networks when they are imposed on a primitive cubic lattice. The resulting artificial structures successfully mimic the geometric and topological features of real proteins, and therefore, provide the opportunity to understand characteristics of protein structures and lead to the creation of synthetic proteins. The proposed method does not involve a chain-fitting step and does not require individual set of reference structures. We start with a cubic lattice, whose lattice sites contain beads representing protein residues. Many cubic lattices are emptied up to 60% vacancy concentration by randomly removing beads while maintaining a connected network of occupied sites. The maximum vacancy concentration of 60% was obtained from first and second nearest neighbor occupancies of real protein residues. A Reverse Monte-Carlo/Simulated Annealing (RMC/SA) simulation that is constrained to fit the average radial distribution function of residues of 278 proteins is then performed. Results indicate that the RMC/SA procedure recovers the average radial distribution function without disturbing other structural properties such as bond orientational order parameters and network topology. Based on various structural properties, our results indicate that these artificially created structures closely resemble real residue networks.