Make genomics project dependency lists data-driven instead of hand-copied

Problem

The list of genomic projects (PlasmoDB, ToxoDB, TriTrypDB, VectorBase, AmoebaDB, CryptoDB, FungiDB, PiroplasmaDB, GiardiaDB, MicrosporidiaDB, TrichDB, HostDB, and now HelminthDB) is hand-copied into 7 separate blocks across 2 files:

Main/lib/xml/workflowTemplates/genomicsRoot.xml — 5 blocks:

• sampleDetailsForEda
• OrthoMCLDependent
• alphaFold
• ExternalResourceUrl
• analyzeWebreadyPartitionedTables

Main/lib/xml/workflow/orthomclGenomics.xml — 2 blocks:

• checkOrthoCoreProteomeDirs dependsExternal list
• checkOrthoPeripheralProteomeDirs dependsExternal list

The only thing keeping them in sync is a comment at genomicsRoot.xml:59:

ATTENTION: This graph has dependsExternals by PROJECT. If you add a project above, you must also add it here

A load-bearing sticky note is the tell that the structure is wrong. Adding HelminthDB (#106) meant 7 edits and a manual audit to confirm nothing was missed; a silently omitted entry produces a workflow graph that runs but quietly skips a project's data.

Proposed direction

The project list is data, not structure. Options, roughly in order of preference:

1. datasetTemplate-driven dependency expansion — the genomicsProject datasetTemplate already enumerates projects to instantiate the per-project subgraphs. Extend ReFlow so a step can declare a dependency on all instances of a template (e.g. <dependsAll template="genomicsProject"/>, and for orthomclGenomics.xml something like <dependsExternalAll template="genomicsProject" step="checkOrthoCoreProteomeDirs"/>). Single source of truth, no duplication anywhere.

2. Shared constant + expansion — declare the list once (<constant name="allGenomicsProjects">) and have the XML expand it into depends / dependsExternal entries. Less invasive than (1), still one place to edit.

3. Validation-only fallback — if neither expansion mechanism is worth building, add a test/lint step that asserts all 7 blocks contain the same project set. Doesn't remove the duplication but converts a silent skip into a loud failure.

Option 1 also makes the naming convention (${projectName}_checkOrthoCoreProteomeDirs) explicit in the graph rather than implied by string concatenation in two places.

Acceptance

Adding project #14 should be a one-line change in one file, with no cross-file audit required.

🤖 Generated with Claude Code

Start by reading the seven dependency blocks in Main/lib/xml/workflowTemplates/genomicsRoot.xml and Main/lib/xml/workflow/orthomclGenomics.xml, then trace how ReFlow expands datasetTemplate dependencies. Compare the proposed template-driven and constant-based approaches, and inspect existing workflow tests or validation commands if present. Done means adding project #14 requires one line in one file without a cross-file audit.