CryoETSim: simulated cryo-electron tomograms for self-supervised representation learning with the CryoSiam framework

Deposition Overview

The CryoETSim deposition provides 400 simulated cryo-electron tomograms generated with the cryo-TomoSim (CTS) simulator for the CryoSiam self-supervised representation-learning framework (Stojanovska et al., bioRxiv 2025.11.11.687379). It spans three scenes: a general macromolecular-crowding scene (300 runs, 141 PDB-derived complexes plus cytoskeletal filaments and membrane proteins), a chromatin scene (50 runs of nucleosome arrays and DNA filaments), and a minimal-bacterium scene (50 runs of bacterial ribosomes, RNA polymerase, membrane proteins, and DNA filaments). Each run provides four reconstructed tomogram variants and paired voxel-level ground-truth masks: a coarse 5-class semantic mask and a fine, scene-specific semantic mask (149/64/88 classes for the general/chromatin/minimal-bacterium scenes); the general scene additionally includes per-particle instance masks with coordinates and volumes.