feat: enrich RO-Crate output files with EDAM format IDs via tataki

[inutano]

Summary

• Adds add_tataki_edam() to sapporo/ro_crate.py, called at the end of generate_ro_crate_metadata() before final JSON-LD serialisation
• Runs tataki via Docker against all files in outputs/, parses the JSON result, and replaces each File entity's encodingFormat with a proper EDAM ontology entity
• Enables tonkaz Level 1–3 file-content comparison on typical nf-core pipeline outputs

What changes

Before (encodingFormat was a mixed array):

"encodingFormat": ["text/tab-separated-values", {"@id": "https://www.iana.org/..."}]

After (proper EDAM entity, as tonkaz expects):

"encodingFormat": {"@id": "http://edamontology.org/format_3475"}

tataki detects formats by file content (not just extension), covering TSV, CSV, JSON, HTML, PDF, PNG, SVG in addition to the existing genomic formats (BAM, VCF, FASTQ, …).

Design

• Separate images: tataki runs as its own Docker container; no changes to the sapporo image
• Best-effort: if Docker is unavailable or tataki fails, a warning is logged and the crate is left unchanged — the run never fails because of enrichment
• Replaces only when detected: files tataki cannot identify keep their existing encodingFormat
• Single batch call: all output file paths are passed to tataki in one docker run invocation

Dependency

Requires ghcr.io/sapporo-wes/tataki:latest to be available on the Docker host. This image is built and published via sapporo-wes/tataki#16.

Test plan

• Run nf-core/rnaseq via Sapporo and verify ro-crate-metadata.json contains encodingFormat: {"@id": "http://edamontology.org/format_XXXX"} for TSV, HTML, JSON output files
• Verify tataki failure (image not present) does not break run completion or crate generation
• Run tonkaz on two identical runs and confirm Level 3 score > 0

🤖 Generated with Claude Code