Toward de novo design of bioadhesives with classical DFT and genetic algorithm

Sea animals like mussel and the sandcastle worm can adhere to wet surfaces by using a collection of flexible proteins. A good understanding of these adhesives may result in novel concepts (viz. biomimicry) to solve the need for improved adhesives that can function well in water. Current research into bioadhesive materials is mostly focused on identifying the amino acid sequence of polypeptides and understanding their contribution towards the overall adhesion strength. A sequence specific coarse-grained model can capture important parameters include electrostatic charge, chain length, ordering, and various modifications of the natural amino acids (e.g., from tyrosine to Dopa). By integrating the classical density functional theory (DFT) and machine-learning strategies and generic algorithms, we investigate the bioadhesive behavior of polypeptides under various environmental conditions and validate the theoretical results with known experimental data. The combination of molecular modeling with machine-learning strategies allows us to sample the polypeptide sequence and identify conditions leading to the maximum adhesive strengths.