Project Summary/Abstract Cryogenic electron-microscopy (Cryo-EM) is a new and rapidly evolving methodology for structural biology. Its success is due to advances in Direct Detector Cameras and new algorithms. This proposal seeks to develop the next generation of algorithms focused on reconstructing the structure of small molecular weight proteins, proteins with multiple conformational states, and developing algorithms that explain why resolution limits occur in Cryo-EM. The proposed research seeks to address important bottlenecks in the above problems. First, we seek to find useful statistics to detect the so-called “Einstein-from-noise” problem. This problem affects the reconstruction of small molecular weight proteins, giving catastrophically bad reconstructions. We propose to investigate this phenomenon theoretically and computationally, so that it can be detected reliably. Second, we seek to develop the next generation of algorithms for reconstructing multiple conformational states of a heterogeneous protein. In particular, we seek algorithms which are stable and which give high resolution reconstructions that can be organized in biologically meaningful ways. Finally, we seek to explore a new direction in Cryo-EM algorithms. We propose to develop a methodology which explains what causes resolution limits in a Cryo-EM reconstruction. Current methods can estimate the resolution of a reconstruction without providing any rationale for its cause. The new methodology will enable users to use data and algorithm settings more effectively. The proposed research builds on the P.I. previous NIGMS supported work on single particle reconstruction in Cryo-EM.

Project Narrative Cryogenic electron-microscopy (Cryo-EM) is a new and rapidly evolving methodology for reconstructing the three-dimensional structure of proteins and biological macromolecules. The next generation of Cryo-EM problems face bottlenecks in reconstructing small molecular weight proteins, heterogeneous proteins with many conformational states, and in creating explanations for limited resolutions. This research proposes mathematical and algorithmic solutions to these problems.