We used cellular automata (CA) for the modeling of the temporal folding of proteins. Unlike the focus of the vast research already done on the direct prediction of the final folded conformations, we will model the temporal and dynamic folding process. To reduce the complexity of the interactions and the nature of the amino acid elements, lattice models like HP were used, a model that categorizes the amino acids regarding their hydrophobicity. Taking into account the restrictions of the lattice model, the CA model defines how the amino acids interact through time to obtain a folded conformation. We extended the classical CA models using artificial neural networks for their implementation (neural CA), and we used evolutionary computing to automatically obtain the models by means of Differential Evolution. As the iterative folding also provides the final folded conformation, we can compare the results with those from direct prediction methods of the final protein conformation. Finally, as the neural CA that provides the iterative folding process can be evolved using several protein sequences and used as operators in the folding of another protein with different length, this represents an advantage over the NP-hard complexity of the original problem of the direct prediction.

We used cellular automata (CA) for the modeling of the temporal folding of proteins. Unlike the focus of the vast research already done on the direct prediction of the final folded conformations, we will model the temporal and dynamic folding process. The CA model defines how the amino acids interact through time to obtain a folded conformation. We employed the HP model to represent the protein conformations in a lattice, we extended the classical CA models using artificial neural networks for their implementation, and we used evolutionary computing to automatically obtain the models by means of Differential Evolution. Moreover, the modeling of the folding provides the final protein conformation.

In the difficult ab initio prediction in protein folding only the information of the primary structure of amino acids is used to determine the final folded conformation. The complexity of the interactions and the nature of the amino acid elements are reduced with the use of lattice models like HP, which categorizes the amino acids regarding their hydrophobicity. On the contrary to the intense research performed on the direct prediction of the final folded conformation, our aim here is to model the dynamic and emergent folding process through time, using the scheme of cellular automata but implemented with artificial neural networks optimized with Differential Evolution. Moreover, as the iterative folding also provides the final folded conformation, we can compare the results with those from direct prediction methods of the final protein conformation.