Computational Developments and Applications for Cryo-Electron Tomography

Cellular organelles and biological macromolecules such as proteins play a fundamental role in almost all life sciences. Structural biology studies the molecular structure of these particles to gain information about their morphology and function. A quickly evolving technique in this field is cryo electron-tomography (cryo-ET). Its big advantage is the ability to determine the three dimensional structure of particles in their native environment. A powerful method for image analysis in cryo-ET is subtomogram averaging (STA), where repetitions of the same particle of interest within a tomogram are computationally extracted, aligned to a common reference and averaged. This procedure significantly increases the signal of the underlying structure. A main bottleneck in cryo-ET and STA is the low throughput in sample preparation, data acquisition and image processing. With the goal to overcome this challenge, we developed computational methods to automate and streamline key steps in image processing. Two projects were designed for this purpose: In the first project, we streamlined the complete processing pipeline of STA and minimized the manual user interaction. In the second project we developed a new algorithm to fully automate the alignment of tomographic tilt series. We further developed a range of specialized tools for STA. We applied them on biological samples in combination with our established methods for increased throughput. Thereby we determined the architecture of protein coats involved in membrane trafficking and shed light on the underlying mechanics. With our new methods we successfully reduced processing times for STA, answered biological questions and laid the foundation for further developments