Published February 16, 2026 | Version v1.0.0
Dataset Open

Integrated Human Early Embryo Atlas with Temporal Annotations

  • 1. ROR icon Wuhan University

Description

This repository provides the Human Early Embryogenesis Atlas used in the Gene-Chronos study, “Gene-Chronos: developmental time inference from single-cell transcriptomes using a pre-trained Geneformer adapter”.

The atlas is a harmonized resource compiled from 12 publicly available human embryo single-cell and spatial transcriptomic datasets spanning early human development from Day 0 to Day 29 (E0–E29), comprising approximately 380,000 cells.

This dataset was constructed to support biological time inference, temporal reconstruction, and cross-dataset transfer evaluation in early human embryogenesis. In the accompanying manuscript, it is used as a large-scale atlas resource for evaluating developmental time prediction and transfer generalization.

The processed data are provided in unified .h5ad format for downstream computational analysis.

Files included:
- Integrated_Human_Embryo_Atlas.tar.gz: compressed archive containing the processed .h5ad files.

Detailed dataset sources, accession numbers, developmental stages, and composition summaries are provided in the accompanying manuscript and supplementary materials.

This dataset was generated to support the Gene-Chronos study. Users of this resource are encouraged to cite the associated Gene-Chronos publication when it becomes available.

 

Technical info (English)

AnnData structure

File format:
- AnnData (.h5ad)

Dimensions:
- Cells (n_obs): 387,062
- Genes (n_vars): 15,691

obs (cell metadata) columns:
- dataset
- time_numeric
- celltype
- time
- day
- dataset_label
- donor
- cell_type
- title
- species
- n_genes
- celltype_meta
- celltype_rf

var (gene metadata) columns:
- n_cells

Additional slots:
- obsm: none
- layers: none
- uns: none

Files

Files (1.4 GB)

Name Size Download all
md5:50a716f8f160dcb0bdef0c2a2c3f55c4
1.4 GB Download