search-ena

name	search-ena
description	Search the EMBL-EBI European Nucleotide Archive (ENA) for sequencing studies, runs, samples, and experiments using the ENA Portal REST API.

Skill: Search ENA

Use When

• A user wants to find sequencing data in the European Nucleotide Archive (ENA) hosted at ebi.ac.uk.
• A user needs to look up ENA/SRA accessions: PRJEB/PRJNA (study), ERS/SRS (sample), ERX/SRX (experiment), ERR/SRR (run).
• A user wants to query ENA by organism, tissue, assay type, instrument, library strategy, or free-text keyword.
• A user wants structured metadata (TSV/JSON) for runs or samples in a study without installing any software.
• A user needs FASTQ or BAM download URLs (FTP or HTTPS) for ENA runs.
• A user wants to find public datasets equivalent to those in NCBI SRA (ENA mirrors most SRA submissions).

Inputs

• Required:
  • one of: ENA/SRA accession (PRJEB/PRJNA/ERR/SRR/ERS/SRS/ERX/SRX), free-text search query, or taxonomy ID
• Optional:
  • result type (ENA entity): study, sample, experiment, run, analysis (default: run)
  • organism filter (scientific name or NCBI taxonomy ID)
  • library strategy filter (e.g., RNA-Seq, WGS, AMPLICON, ChIP-Seq)
  • instrument platform filter (e.g., ILLUMINA, OXFORD_NANOPORE, PACBIO_SMRT)
  • library layout: PAIRED or SINGLE
  • fields to return (comma-separated ENA field names, or all for full metadata)
  • date range: first_public range in YYYY-MM-DD
  • maximum results (limit, default 100)
  • output format: tsv or json
  • output file path

Workflow

1. Construct an ENA Portal API search request:

   GET https://www.ebi.ac.uk/ena/portal/api/search
     ?result=<result-type>
     &query=<lucene-query>
     &fields=<field-list>
     &format=<tsv|json>
     &limit=<n>
   

   • For accession-based lookup: query=study_accession="<PRJEB>" or query=run_accession="<ERR>"
   • For text search: query=<term> with optional field qualifiers such as scientific_name="Homo sapiens"
   • For taxonomy: query=tax_eq(<taxon_id>) or query=tax_tree(<taxon_id>) (includes descendants)

2. Recommended field sets by result type:

   • run: run_accession,experiment_accession,sample_accession,study_accession,scientific_name,instrument_platform,library_strategy,library_layout,read_count,base_count,fastq_ftp,fastq_bytes,first_public
   • study: study_accession,secondary_study_accession,study_title,tax_id,scientific_name,study_description,first_public
   • sample: sample_accession,secondary_sample_accession,scientific_name,tax_id,sample_title,collection_date,country,tissue_type

3. Parse the response to build a summary table.

4. Extract fastq_ftp or fastq_aspera columns for direct download links.

5. For large queries (>1000 runs), use limit and offset for paging, or stream with format=tsv.

6. Report download commands using wget or curl for the FTP URLs.

Output Contract

• API request URL used
• Number of matching records
• Summary table with key fields per result type (accession, organism, strategy, platform, read count, size, date)
• FASTQ download URLs (FTP and HTTPS) for run-level queries
• Saved output files when a path is provided:
  • results.tsv or results.json
  • fastq_urls.txt (one URL per line, ready for wget -i or aria2c -i)

Limits

• This skill depends on live access to https://www.ebi.ac.uk/ena/portal/api/.
• ENA Portal API returns at most 100 000 rows per request; use paging for larger result sets.
• Not all SRA submissions are mirrored in ENA immediately; newly submitted data may not be available yet.
• Controlled-access datasets (e.g., from EGA) are not accessible through the ENA Portal API.
• The fastq_ftp field may be empty for runs that are only available as SRA format; in that case fall back to sra_ftp or use fasterq-dump.
• ENA Lucene query syntax differs from NCBI E-utilities query syntax; do not mix them.
• Common failure cases:
  • using NCBI-style [field] tag syntax instead of ENA field="value" syntax
  • requesting result=run but filtering by study-level fields not propagated to run records
  • fastq_ftp returning multiple semicolon-separated URLs (for paired-end runs); must split on ;
  • omitting tax_tree() and only getting exact taxon matches, missing subspecies/strains