Submitting to ENA

SeqDesk automates the submission process to the European Nucleotide Archive. This page covers the step-by-step submission workflow.

Submission Process

XML Generation

SeqDesk generates ENA-compliant XML for each entity type:

Study XML

<PROJECT_SET>
  <PROJECT alias="my-study-alias">
    <TITLE>Study Title</TITLE>
    <DESCRIPTION>Study description...</DESCRIPTION>
    <SUBMISSION_PROJECT>
      <SEQUENCING_PROJECT/>
    </SUBMISSION_PROJECT>
  </PROJECT>
</PROJECT_SET>

Sample XML

<SAMPLE_SET>
  <SAMPLE alias="HG001">
    <TITLE>Human Gut Sample 1</TITLE>
    <SAMPLE_NAME>
      <TAXON_ID>408170</TAXON_ID>
      <SCIENTIFIC_NAME>human gut metagenome</SCIENTIFIC_NAME>
    </SAMPLE_NAME>
    <SAMPLE_ATTRIBUTES>
      <SAMPLE_ATTRIBUTE>
        <TAG>geographic location (country and/or sea)</TAG>
        <VALUE>Germany</VALUE>
      </SAMPLE_ATTRIBUTE>
      <!-- Additional MIxS fields -->
    </SAMPLE_ATTRIBUTES>
  </SAMPLE>
</SAMPLE_SET>

ENA API Endpoints

Requests use HTTP Basic authentication with your Webin credentials and multipart form data containing the XML files.

Connection test

The Test Connection button in ENA settings does not register anything: it sends a minimal multipart POST to the drop-box endpoint with a VALIDATE action and a throwaway project. A success="true"/success="false" receipt means the credentials authenticated; a 401 means they did not. The username must match Webin-12345 (the ^Webin-\d+$ pattern) before the request is even sent.

Broker-account mode

If the install runs as an ENA broker (configured in ENA settings), the registration path adds a center_name attribute — taken from the configured center name — to the PROJECT and SAMPLE XML it generates, so records are attributed to the brokering center. Broker mode applies to the XML registration path; the submg pipeline below does not inject a center name.

Submission Tracking

Each study registration creates one Submission record (the XML path creates study submission records; it does not create separate per-sample records):

There is no separate receiptXml, requestPayload, or error-message column — the receipt and error text live inside the response JSON. See the Submissions Dashboard for the full record shape and the status lifecycle.

Resubmitting

The XML path does not batch or retry samples individually. If a registration fails or partially succeeds, fix the underlying records and re-run it from the Submissions Dashboard — “Retry” simply re-POSTs /api/admin/submissions for the same study, regenerating fresh XML. When a study already has an accession, the re-run skips re-registering the project and submits only the samples that still lack an accession.

Submitting reads and assemblies with submg

The XML path above registers the study and samples (metadata) only. To submit the actual sequencing data — reads, the assembly, and optional bins — SeqDesk runs the submg pipeline (“Submit to ENA”), which wraps the submg  CLI. It runs as a normal pipeline run: it is visible to users but admin-triggered (userCanStart is false), can be scoped to a whole study or a subset of samples, and writes accessions back to the database when it finishes.

Prerequisites

Per sample, submg requires:

• Paired-end reads — each read must have both an R1 and an R2 FASTQ; submg rejects samples with no paired-end read files.
• MD5 checksums on both R1 and R2.
• An assembly FASTA that exists on disk (run the MAG pipeline first if a sample has no assembly).
• A study ENA accession (PRJ...) — the study must already be registered via the XML path.
• Checklist metadata — collection date and geographic location (country and/or sea) are required; other checklist fields are passed through.

When the ENA target is the Test server, the study’s test registration must also be less than 24 hours old, because ENA expires test registrations.

An assembly coverage value is resolved from sample or order custom fields (coverage_depth, coverage_value, target_coverage, and similar keys). If none is found it defaults to 1 and the run emits a warning so you know the coverage was assumed rather than measured.

What it submits

submg builds a YAML config per sample and calls submg submit with --submit_samples --submit_reads --submit_assembly (and --submit_bins when bins are present and enabled). When it completes, SeqDesk parses the submission logs and writes back:

• sampleAccessionNumber (ERS...) and biosampleNumber (SAMEA...) on each sample
• run and experiment accessions on each read
• the assembly accession (ERZ...)
• bin accessions when bins were submitted

This BioSample (SAMEA...) write-back is unique to the submg path.

Test vs Production