UCSF

Proteins perform most of biological functions. The ultimate approach to understand protein functions is designing novel protein functions from scratch. The ability to design novel functional proteins would also be revolutionary for biology research, medicine and synthetic biology. Proteins function by placing specific chemical groups at precise 3-dimensional (3D) positions and orientations. Therefore, de novo functional protein design relies on precise control of protein 3D structures. In this thesis, I review the recent advances in de novo protein design (chapter 1). Then I describe two of my research projects that aim to extend the ability of computational methods to design precise protein 3D structures. In the first project (chapter 2), I developed a method termed loop-helix-loop unit combinatorial sampling (LUCS) that systematically samples geometries of protein secondary structures. I applied LUCS to designs de novo protein fold families of two topologies. In the second project (chapter 3), I developed two computational methods termed fragment kinematic closure (FKIC) and loophash kinematic closure (LHKIC) for modeling protein local segments. FKIC predicts protein local segment structures with significantly higher accuracy and efficiency than previous methods. LHKIC designs local segment structures by combining the power of the loophash algorithm and the kinematic closure algorithm. I expect these new methods can bring the ideal of designing arbitrary protein functions closer to reach.