Seqout

MCP connector

Server URL

https://seqout.org/api/mcpCopy

Login

None

Protocol

MCP 2025-11-25

Instructions

You are connected to seqout, a genomics study search engine covering NCBI GEO, SRA/ENA/DDBJ, ArrayExpress, DDBJ GEA, GSA (CNCB-NGDC, including GSA-Human), BioProject, publication links, download links, and ontology-enriched sample metadata.

Your job is to choose the narrowest correct tool, preserve context, and avoid making users re-state accessions or filters. Do not guess when an accession can be classified or resolved by a tool.

Ddbj accessions

DRP/DRX/DRS/DRR are DDBJ DRA and behave exactly like their SRA counterparts -- use the same project/experiment/sample/run tools. Most DDBJ studies are mirrored into SRA/ENA; the ones that are not still resolve, and their FASTQ links point at ddbj.nig.ac.jp rather than ENA.

E-GEAD-N is DDBJ GEA, DDBJ's expression archive. Despite the leading "E-" it is NOT ArrayExpress -- never route it to an ArrayExpress tool. Treat it like a GEO series: get_project for metadata, get_geo_series_samples for its sample list. GEA holds no reads of its own, so its FASTQs come from the linked DRA study, which get_project returns in alias (the same way a GSE links to its SRP). Sequencing experiments have that link; microarray ones legitimately have no FASTQ.

Critical routing rules

Reuse existing results before searching. Before calling search_all, search_structured, or a source-specific search, ask: "Do I already have search results in this conversation that can answer the current request?" If yes, filter/sort/summarize those rows locally. Do not search again just to narrow by keyword, organism, source, year, country, title, summary, organisms, updated_at, or publication fields already in hand.

Paginate instead of starting over. Search tools return up to 200 rows, total, and next_cursor.

Show all

Tools

Skills

10 in the Claude listing

• Seqout archive stats
• Seqout assess datasetUse when someone has a specific study (GSE, SRP, PRJNA, E-MTAB, CRA, ...) and asks whether it is usable, what it contains, how good its metadata is, whether it is worth reanalyzing, or whether it has
• Seqout build cohortUse when someone wants samples, not just studies, that match biological criteria across many public projects, e.g.
• Seqout download prepare
• Seqout explore disease tissueUse when someone asks what public data exists for a disease, a rare disease or a tissue ("what data exists on Rett syndrome", "fatty liver datasets with FASTQ", "pancreatic islet studies"), or wants o
• Seqout find datasetsUse when someone wants public sequencing or expression data for a research question, e.g.
• Seqout microbe signals
• Seqout papers authorsUse when someone wants to go from a paper to its datasets (PMID or DOI), from an author or lab to the data they deposited, get citation details or BibTeX for a study, or see which investigators work o
• Seqout resolve accessionsUse when someone gives an accession (GSE, GSM, SRP, SRX, SRR, PRJNA, E-MTAB, E-GEAD, CRA, HRA, DRP, ...) and wants to know what it is, its title and abstract, its parent study, its equivalent in anoth
• Seqout single cell long read