Natural protein motifs and their use in design

Though the fundamental forces that drive protein folding and function are fairly well understood, calculations involving proteins are difficult. With a large number of atoms and bonds, most proteins structures have many degrees of freedom manageable only through some combination of physical approximations and probability-based expectations. Recent years have offered a dramatic increase in the number of known protein structures, empowering unbiased computational approaches to perform the detection of structural motifs and to quantitatively relate local structure to global structure and amino acid sequence. In this work, particular attention is paid to α-helices. First, the sequences of alpha helices are evaluated relative to their termini. Sequences are shown to include periodic propensities at depths hitherto unstudied, and periodicities from opposite termini are argued to explain the periodicity in preferences for different helix lengths. Second, a survey of membrane helices is presented, including a novel study of proline kinks. This is followed by a comprehensive search for motifs in loops that covalently link pairs of α-helices. While some linking motifs have been identified by previously through manually inspection of smaller datasets, our approach exhaustively parameterizes all links in a modern database automatically. Structural motifs are then shown to have distinct preferences for interhelical geometries and physiochemical composition. To test these links, a successful lab experiment to connect helices was performed and is explained. Knowing which motif is appropriate in a given context does not give exact coordinates to use in a designed structure, so a novel approach for protein design, called Stitch, was developed and successfully applied to several design projects. Candidate stitches, each a natural protein fragment, are selected based on RMSD to the target and on their similarity to known motifs. While protein fragments have been used in design and structure prediction previously, Stitch is distinct in using larger fragments and requiring a larger overlapping region of backbone. Stitch\u27s success inspired another project, Sisyphus, which builds infinite scaffolds based on nature\u27s agreement both at stitching sutures and interfacial contacts. Sisyphus is applied to the design of a triangular prism scaffold with α-helix legs