CryoAtlas: A Large Curated Dataset and Unified Benchmark for Cryo-EM Atomic Model Building (Train Part 4)
Year: 2026
Type: dataset
Abstract: This dataset is Part 4 of 4 of the CryoAtlas v1.0 training set for research on automated protein atomic model building from cryo-EM density maps. The complete training set contains 16,657 curated entries derived from EMDB and PDB records available up to March 2026. Following map–model alignment, sequence mapping, and quality filtering, each entry provides paired experimental density maps, atomic structures, and protein sequences.This part is organized by complete EMD entries and includes density maps, corresponding atomic structures in CIF format, and original, aligned, and terminal-trimmed protein sequences in FASTA format. The four parts jointly constitute the fixed training split and contain no overlapping entries.This dataset belongs to the CryoAtlas v1.0 Package, which brings together all four training-set parts, the complete test set, and benchmark results. The package provides a unified access point for data retrieval, method development, and reproducible evaluation.
Source: ScienceDB
Author: 李明瑞
Institution: ShanghaiTech University
Language: English
Subfield: Structural Biology
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7223258878