Skill: Search GSA

Use When

• A user wants to find sequencing data deposited in the CNCB Genome Sequence Archive (GSA) at ngdc.cncb.ac.cn.
• A user needs to look up GSA accessions: CRA (project), CRX (experiment), CRR (run), CRS (sample), or SAMC (biosample).
• A user wants to retrieve run-level metadata (read count, file size, platform, library strategy) for data in GSA.
• A user wants to search GSA by organism, tissue, disease, keyword, or data type.
• A user wants to list all runs belonging to a GSA project (CRA accession).
• A user needs the download URL or FASP path for CRR files from CNCB.

Inputs

• Required:
  • one of: free-text search query, GSA accession (CRA/CRX/CRR/CRS/SAMC), organism name, or NCBI taxonomy ID
• Optional:
  • data type filter: Genome, Transcriptome, Metagenome, Epigenome, Variation, Proteome
  • library strategy filter (e.g., RNA-Seq, WGS, ChIP-Seq, ATAC-Seq)
  • sequencing platform (e.g., Illumina, PacBio, ONT)
  • organism filter (scientific name or taxonomy ID)
  • output format: json or tsv
  • maximum results (size, default 20)
  • output directory

Workflow

1. Determine the access method:
   • Direct accession lookup: construct the CNCB accession detail URL:
     • https://ngdc.cncb.ac.cn/gsa/browse/<CRA-accession> for projects
     • https://ngdc.cncb.ac.cn/gsa/run/<CRR-accession> for individual runs
   • Keyword search: use the GSA search API endpoint:

     GET https://ngdc.cncb.ac.cn/gsa/search?searchTerm=<query>&page=1&pageSize=20

   • Programmatic metadata fetch: use the GSA data API:

     GET https://ngdc.cncb.ac.cn/gsa/api/v1/project/<CRA>
     GET https://ngdc.cncb.ac.cn/gsa/api/v1/run/<CRR>

2. Parse the response to extract: accession, title, organism, submission date, run count, total bases, platform, and library strategy.
3. For project-level queries, enumerate all CRR runs and collect per-run metadata.
4. Construct download URLs for CRR files:
   • FTP: ftp://download.cncb.ac.cn/gsa/<CRA>/<CRX>/<CRR>/
   • HTTPS: https://download.cncb.ac.cn/gsa/<CRA>/<CRX>/<CRR>/
5. Report a summary table and the accession list.

Output Contract

• Search or fetch request URL used
• Number of matching projects or runs
• Summary table with columns: Accession (CRR/CRA), Title, Organism, Platform, Strategy, Spots, Bases, Size, Submission Date
• Download URL(s) for CRR files
• Accession list saved to accessions.txt when an output directory is provided
• Optional metadata.json or metadata.tsv with full per-run fields

Limits

• GSA web API is subject to change; confirm endpoint availability with a test request if needed.
• Some datasets in GSA are access-controlled and require registration at ngdc.cncb.ac.cn before download.
• GSA accessions follow CNCB conventions (CRA/CRX/CRR/CRS) distinct from NCBI (SRP/SRX/SRR/SRS); do not mix them.
• CNSA (China National Sequence Archive, for assembled genomes) is a separate archive also at CNCB; GSA is specifically for raw sequencing data.
• The FTP server may be slow for users outside mainland China; HTTPS is generally preferred.
• Common failure cases:
  • confusing CRA (project) with CRR (run) accession types
  • attempting to use NCBI E-utilities with GSA accessions (they are not indexed by NCBI)
  • missing authentication cookie when accessing controlled-access datasets
  • constructing download paths without the correct CRA/CRX/CRR hierarchy