Decide the Galaxy-side shape of external reference data declared by a Nextflow pipeline.
Directory Mold with only index.md frontmatter.
source-specific fields are coherent.
eval.md not written yet.
4 typed references; 0 resolver issues.
All on-demand references describe triggers.
Hypothesis references include verification.
Typed Mold references describe what casting consumes and when the generated skill should load each artifact.
Structured contract copied for validation or lookup.
Background synthesis loaded by explicit progressive-disclosure metadata.
Background synthesis loaded by explicit progressive-disclosure metadata.
Background synthesis loaded by explicit progressive-disclosure metadata.
Reviewable Markdown brief: per-asset reference inputs, requiredness, rebuild-on-absence behavior, iGenomes key disposition, open questions.
Structured Nextflow pipeline summary emitted by [[summarize-nextflow]]; the source for reference-data param evidence.
Classify the ways the Nextflow pipeline uses reference data using nextflow-reference-data-classification, then use nextflow-to-galaxy-reference-data-mapping to determine and describe how each classified asset maps onto the target Galaxy workflow. Output is a reviewable Markdown brief consumed by nextflow-summary-to-galaxy-interface, nextflow-summary-to-galaxy-data-flow, and nextflow-summary-to-galaxy-template before they pin workflow inputs, data-flow edges, and the gxformat2 skeleton.
The summary-nextflow schema surfaces reference-data evidence directly — read these fields before falling back to raw workflow.conditionals[] or guard-text inference:
reference_assets[] — the curated asset inventory. Each entry has param (FK into params[]), asset_kind (fasta, fasta_index, sequence_dictionary, bwa_index, tabix_index, gtf, gff, bed, vcf, database, other), required, source_kind, used_by (consuming subworkflow/process names), and evidence.confidence. Use this as the primary input to classification — it already filters out non-asset path params and attributes callers.reference_rebuilds[] — detected compute-if-missing branches. Each entry binds asset_param to a builder process plus its builder_outputs, the verbatim guard, resolved guard_params, and fallback_for (the <asset>_in-style take when paired). When reference_rebuilds[] is non-empty, the compute-if-missing classification applies to those asset_param values. Empty reference_rebuilds[] does NOT mean “no compute-if-missing” — some pipelines use positive-form idioms (if (<asset>) { unpack } else { build }) that the detector does not yet match (see jmchilton/foundry#229 follow-ups).params[].source_kind == "getGenomeAttribute" plus params[].source_expression — the key-expanded bundle signal. A pipeline that synthesizes many path params via getGenomeAttribute('attr') is using the iGenomes pattern; classify the bundle accordingly.subworkflows[].invocations[] — caller-side argument binding onto take: names. Use when reference_assets[].used_by is ambiguous or when verifying that the same param threads into multiple subworkflows.If a needed signal is missing from the summary (e.g. RNA-seq’s positive-form rebuild idiom is not yet detected), say so in the brief and flag the asset as rebuild-unverified rather than re-parsing source files.