Computational protein and peptide design encompasses the creation of novel proteins from scratch or the modification of natural proteins, and seeks to identify the optimal amino acid sequence to adopt a specified fold or function. The resulting engineered protein variants can provide insights into the biophysical properties of proteins or be formulated as diagnostic tools, therapeutic agents, or functional biomaterials.

In contrast to conventional protein design strategies such as targeted evolution that construct novel sequences through near-random mutagenesis, computational protein design constructs novel sequences through a combination of biochemical intuition as well as data that reflect our current understanding of proteins in general and for specific folds. For example, in our lab, we have used MD simulations to construct dynamic libraries of amino acid propensities that reflect the intrinsic conformational preferences for the backbones and sidechains of both L- and D-amino acids. Reference MD simulations of protein designs, fold-family-specific sequence trends, and broader insights from the Dynameomics database also inform protein design efforts.

(Left) α-sheet conversion, conformational properties, and peptide designs. (A) β- to α-sheet conversion of transthyretin. At 0 ns (top left) the residues of interest form a β-hairpin. After 30 ns (top right) the β-sheet has converted to an α-sheet. (B) Intrinsic residue propensities for L- and D-alanine were calculated from 100 ns of MD simulations of a GGXGG peptide system. (C) Peptide designs are single turn hairpins, with the exception of α3, which contains a cyclic peptide backbone resulting in two turns. (image adapted from DOI)