Research Projects

Cryo-electron subtomogram alignment and clustering have become a primary method for resolving protein and complex structures in recent years. Most cryo-EM structures are generated from Single Particle Analysis (SPA), which involves collecting and aligning 2D projections of the target protein to determine the 3D structure. In cryo-electron tomography (cryoET), 3D subtomograms of a protein or a protein complex are extracted and aligned to resolve the 3D structure of the complex. Software has been developed to automate the alignment and clustering workflow for SPA and cryoET. However, most commonly used alignment methods rely on a single scoring function, constrained correlation (CCC) or CCC variants. A recent publication compared over 15 scoring functions for evaluating the quality of 3D subtomogram clusters, showing that there is still potential to use other scoring functions to optimize alignment and clustering. Spectral signal-to-noise ratio-based Fourier Shell Correlation (SFSC) showed the best performance for ranking alignment and contamination errors, even for subtomograms with a low signal-to-noise ratio. To develop new strategies for aligning and clustering cryo-electron subtomograms for resolving high-resolution structures of proteins and protein complexes using SFSC as an optimizing scoring function, Monte-Carlo optimization and machine learning methods guided by the SFSC scoring function will be utilized.