GEO Dataset Search (single-cell / transcriptome)

Describe the data you need in one sentence, e.g. “human liver single cell RNA-seq”, and get the matching GEO series (GSE): title, organism, sample count, release date and whether it is single-cell.

Your search terms are sent by your browser directly to the public NCBI E-utilities service. They do not pass through our servers and we do not log them. Do not enter unpublished project details.

1. Describe the data you need

Try:

Chinese input is understood for common terms only (organisms, tissues, cell types, diseases, assays); English keywords work best.

2. Candidate datasets

Type a description and press “Search”, or click an example above.

Data come from NCBI GEO, searched live through the public NCBI E-utilities API. This page only lists candidates; nothing is downloaded.
How your description is read: a recognised organism (human, mouse…) restricts the organism; “single cell / scRNA-seq / snRNA-seq” requires single-cell wording; RNA-seq or transcriptome restricts to expression profiling by sequencing; other words become keywords, searched together with synonyms (e.g. tumor / cancer / carcinoma).
“Single-cell” is flagged when the title or summary mentions single cell, single nucleus, scRNA-seq, 10x Genomics and similar; a dataset without that wording may still be single-cell, so check its GEO page.
Relevance: keywords in the title count more than in the summary; when you ask for single-cell, datasets without single-cell wording are ranked well down. NCBI’s own ordering is unreliable, so results are re-ranked here.
Sample count is the number of GEO samples (GSM); a single-cell sample usually holds thousands of cells. A SuperSeries bundles several sub-series and its count is the total.
To limit release years, write them in the description: “2018-2020”, “since 2021” or just “2020”. With many matches and no years, only the most recently released part can be fetched.
To be gentle on NCBI, requests are queued (at most 3 per second), so a search takes a few seconds; repeating the same search on this page does not send new requests.