An integrated transcriptomic cell atlas of human neural organoids

Source

PDF: raw/sources/he_2024_an_integrated_transcriptomic_cell_atlas.pdf
Article: Nature
Labs: Barbara Treutlein, J. Gray Camp, Fabian Theis — ETH Zürich / Roche / Helmholtz Munich (joint senior authors)
Status: deep ingested 2026-04-09

Integrated 36 scRNA-seq datasets spanning 26 distinct neural organoid protocols
Total cells: >1.7 million
Reference mapping: against developing human brain cell atlases (4 primary brain references)
Scope: unguided and guided protocols, various brain regions
Deliverables: (1) integrated UMAP atlas, (2) programmatic interface to browse/query, (3) quantitative fidelity scores per protocol per region, (4) diseased organoid annotation framework

Atlas identifies which parts of the human developing brain ARE covered by existing protocols — and which remain uncovered.
Quantifies transcriptomic similarity between primary and organoid cells across all protocols on a common scale.
Identifies primary cell populations that are under-represented in neural organoid models (important gap analysis).
Atlas enables automatic annotation of new organoid datasets and side-by-side comparison of new vs. published protocols.
Disease organoid analysis framework: uses healthy organoid atlas as "control cohort" to find disease-specific deviations.

The single most comprehensive cross-protocol benchmarking resource in the corpus.
Builds directly on Velasco 2019, Bhaduri 2020, Kanton 2019, and many others — provides the meta-analysis they could not do alone.
Answers the "which parts of the brain are we actually modeling" question that no individual protocol paper can answer.
Includes assembloid (cf. Sloan 2018, Meng 2025) and transplantation (cf. Kelley 2024) data where available.

Transcriptomics only — does not integrate electrophysiology, imaging, or morphology across protocols.
Reference primary brain data sets have their own technical artifacts.
"Under-represented" populations may reflect dissociation artifacts rather than absence.
Cross-protocol comparison relies on harmonized annotation, which may mask protocol-specific biology.

The closest thing we have to a definitive cross-protocol mapping of brain region coverage and fidelity.
For any brain region question, this is the first place to check which protocol is best suited.
Developmental stage mapping across protocols provides the maturation-axis quantification that was missing in Round 1.

Velasco 2019 — within-protocol reproducibility that this atlas contextualizes at cross-protocol scale.
Bhaduri 2020 — primary cortex reference data used in the atlas.
Kanton 2019 — one of the foundational datasets integrated into this atlas.
Meng 2025 — perturbation data that can be compared against this atlas as control.

Which brain regions remain most under-represented in existing protocols, and can the atlas guide targeted protocol development?
How does the atlas handle recently published (post-2024) protocols?
Can the disease-organoid framework generalize to non-brain organoid disease models?

These parsed files are helper artifacts. Treat the original raw PDF as the source of truth.