Abstract
Helical symmetry is a common structural feature of many biological macromolecules. However, helical indexing and de novo 3D reconstruction remain challenging. We have developed a computational method, Helicon, which poses helical reconstruction as a linear regression problem with the projection matrix parameterized by the helical twist, rise, and axial symmetry. A sparse search of the twist and rise parameters would allow helical indexing and 3D reconstruction directly from one 2D class average or a raw cryo-EM image. The Helicon method has been validated with simulation tests and experimental cryo-EM images of helical tubes, non-amyloid filaments, and amyloid fibrils. Imaging stitching and L1 regularization of linear regression were shown to improve the robustness for low-twist amyloids and noisy raw cryo-EM images. Using Helicon, we could successfully index the helical parameters and perform de novo reconstruction of a previously unreported, low abundance tau amyloid structure from a publicly available dataset. Graphic Highlights: Helicon enables helical indexing and 3D reconstruction from a single 2D image Formulates helical reconstruction as a linear regression problem Tackles low-twists and raw cryo-EM images with image stitching and regularization Validated with diverse experimental data and a previously unreported tau filament
All Related Versions

This publication has 38 references indexed in Scilit: