Name
Abstract | ||
|---|---|---|
In this work the temporal and dynamic folding of proteins was modeled with neural cellular automata, on the contrary to the ample research performed on the prediction of the final protein structure. Using the Rosetta environment and its coarse-grained representation, starting from an unfolded or partially folded chain, a connectionist model acts like a cellular automaton to define the moves of the dihedral angles of the protein chain. The process is repeated for all the angles of the amino acids and through several time steps until the protein is folded. The neural cellular automaton uses as input information a partial view of the energy landscape, obtained through the consequences in the energy changes when an angle is moved. The neural model learns to decide the best move in each angle in order to minimize the energy of the final folded conformation. The neural cellular automata are automatically obtained by means of differential evolution. Initial results with short proteins are presented and discussed. |
Year | DOI | Venue |
|---|---|---|
2016 | 10.1145/2908961.2931720 | GECCO (Companion) |
Keywords | Field | DocType |
Protein folding, Cellular automata, Evolutionary computing, Differential evolution | Cellular automaton,Protein folding,Computer science,Evolutionary computation,Algorithm,Differential evolution,Artificial intelligence,Energy landscape,Connectionism,Machine learning,Dihedral angle,Protein structure | Conference |
Citations | PageRank | References |
1 | 0.36 | 3 |
Authors | ||
2 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Daniel Varela | 1 | 3 | 2.45 |
| José Santos | 2 | 97 | 14.77 |