Towards unbiased 3D reconstruction in single-particle cryo-electron microscopy

Sammanfattning: Cryo-electron microscopy of freestanding molecules (single-particles) plays a pivotal role in the difficult and pressing challenge of determining the structures of large macromolecular complexes. Molecular volumes are generated by aligning large sets of randomly oriented two-dimensional (2D) projection images in three dimensions (3D) before reconstruction is performed using tomographic techniques. The increasing popularity of the single-particle method is highly correlated with technical advances in instrumentation and computation. This thesis introduces new computational methods for 3D structure determination from electron microscopic projection images of single molecules. The algorithms have been developed to fill a gap in the single particle methodology – the lack of methods for ab initio 3D reconstruction of asymmetrical or low-symmetry molecules co-existing in different functional states. The proposed approach does not rely on a priori information about the structure or the character of the sample heterogeneity, which minimizes template dependence and makes the methods applicable to a wide range of single molecules. The presented algorithms constitute the basis of a new open source software package - SIMPLE (Single-particle IMage Processing Linux Engine). SIMPLE is an efficient and easy-to-use image processing system for semi-automated ab initio 3D reconstruction from challenging single-particle data sets (asymmetrical particles, significant degree of heterogeneity).