Comprehensive semi-supervised automated annotations of macromolecules and organelles in C. reinhardtii tomograms

Deposition Overview

These segmentations of 23 different organelles and macromolecular structures across the 1829 volumes in the C. reinhardtii dataset were generated as a part of a project that explored the scaling of data analyses in cryoET. On the basis of these comprehensive segmentations the dataset was also 'summarized' and represented a searchable database (see method links). All segmentations were generated using convolutional neural networks. Networks to segment macromolecules (membrane, ribosome, F1-F0 complex, RuBisCo complex, chaperonin-containing T complex, and proteasome complex) were prepared using Ais (check method links for network weights). The remaining features were segmented by a single network that was prepared using Pom.

Annotation Methods Summary