Tag Catalog

Tags are how a Foundry’s corpus is browsed. Not how it is typed — that is the type: discriminator and the kind-catalog — but how a reader crosses it: every note carries at least one tag, every tag belongs to a declared facet, and every facet is a closed enum whose members each carry a one-line gloss.

The format is substrate, shared verbatim across instances and specified in Part 4 of the Build with the Astro Stack. The vocabulary is not, and should not be: a domain’s browse axes are the domain’s.

This page renders all three registries as they actually stand, vendored from each instance’s meta_tags.yml.

What the format buys, and why it is worth copying

Three rules do the work, and each exists because its absence caused a specific problem:

Membership is declared, never parsed. A tag is valid because some facet lists it under values — not because its text begins with a facet name. So target/not-a-real-thing is as invalid as nonsense, the slash is a naming convention rather than a rule, and a bare key with no slash is an ordinary member of its facet rather than a documented special case.
Every facet is closed, and every tag has a gloss. There is no open family, no free-form escape hatch. A tag with no gloss is a tag the browse surface cannot document and a reader cannot learn from — so the registry is the complete, permanent catalog of what the corpus can carry, which is what makes a page like this one possible at all.
Browse pages group by the declaring facet. Not by prefix. That is what makes an “other” bucket structurally impossible rather than merely empty today.

A fourth rule is about what tags are for: they are cross-cutting facets only. A note’s kind is never copied into its tags. The first two instances encoded kind-as-tag early, and both removed it — two encodings of one fact drift, and the one in tags is the one nothing validates against.

How to read it

A shared facet key is not a shared meaning. The first two registries both declare topic, and they mean different things — one groups pattern maps, the other sits beneath a domain. The TDA Bioinformatics Foundry does not declare it at all. Facets every instance uses are shown first with their descriptions side by side; partial overlaps remain visible in the per-instance views. That distinction is exactly what otherwise gets lost when someone assumes a key transfers.

The per-instance vocabularies follow in full, with each facet’s own description and every tag’s gloss. A facet declared with no members yet is shown as such — the format permits it while a facet is still being filled, and hiding it would misrepresent the registry.

For the kinds these tags are attached to, see kind-catalog. For what else a new instance inherits versus supplies, see anatomy-of-an-instance.

Generated from each instance's meta_tags.yml. The registry FORMAT is shared and specified in Part 4 of the Build with the Astro Stack; the vocabularies below are each instance's own.

Galaxy Workflow Foundry — 7 facets, 22 tags
TDA Bioinformatics Foundry — 4 facets, 16 tags
Statistical Genomics Foundry — 5 facets, 38 tags

Every facet is a closed enum with a gloss per tag, in every instance. Membership is declared — a tag is valid because a facet lists it, never because its text starts with a facet name — so the slash is a naming convention and an "other" bucket is impossible rather than merely empty.

Facet keys every instance uses

A shared KEY is not automatically a shared meaning. Read every description before assuming a tag transfers — this is exactly where a vocabulary silently diverges.

`meta`

Galaxy Workflow Foundry — Meta: Foundry-meta notes — about the Foundry's own tooling and casting system, or an external harness/tool it evaluates. The one facet whose member carries no slash.; meta
TDA Bioinformatics Foundry — Meta: Foundry-meta notes — about the Foundry's own design, machinery, or casting system rather than about topological data analysis. The one facet whose member carries no slash.; meta
Statistical Genomics Foundry — Meta: Foundry-meta notes — about the Foundry's own design, tooling, or casting system rather than about statistical genomics. The one facet whose member carries no slash.; meta

Vocabularies

Galaxy Workflow Foundry

meta Meta

Foundry-meta notes — about the Foundry's own tooling and casting system, or an external harness/tool it evaluates. The one facet whose member carries no slash.

meta: Foundry-meta note — about the Foundry's own tooling, casting system, or an external harness/tool it evaluates

cli CLI not universal

CLI affiliation — every cli-tool and cli-command note carries one. Drives per-tool browse pages and action-Mold reference surfaces.

cli/gxwf: gxwf CLI (Galaxy workflow design-time tooling)
cli/galaxy-tool-cache: galaxy-tool-cache CLI (fetch/cache/inspect Galaxy tool metadata)
cli/planemo: Planemo CLI (Galaxy workflow runtime testing)
cli/cwltool: cwltool reference runner / validator for CWL
cli/cwl-utils: cwl-utils CLI (cwl-normalizer and friends)
cli/foundry: Foundry-shipped CLI bins (validators and harness tooling)

prompt Prompt not universal

Reusable upstream or Foundry-authored prompt families.

prompt/galaxy-internal: Prompt sourced from Galaxy's internal agent prompt library

source Source not universal

What shape of input a Mold or source-pattern consumes.

source/paper: Mold consumes paper-shaped input
source/interview: Mold or pipeline starts from a free-form user interview
source/freeform: Mold consumes a normalized free-form source summary
source/nextflow: Mold consumes Nextflow-shaped input
source/cwl: Mold consumes CWL-shaped input
source/galaxy: Mold consumes an existing Galaxy gxformat2/.ga workflow as its source

target Target not universal

What system a Mold produces for.

target/galaxy: Mold targets Galaxy
target/cwl: Mold targets CWL

tool Tool not universal

Which CLI surface a Mold wraps. Distinct from `cli/*`, which says what a cli-tool or cli-command note is *about*.

tool/planemo: Mold wraps Planemo CLI surface

topic Topic not universal

Foundry-authored pattern/MOC topics — the subject maps patterns group under.

topic/galaxy-transform: Galaxy data-shape transformation pattern maps
topic/collection-transform: Galaxy collection transformation pattern map
topic/tabular-transform: Galaxy tabular transformation pattern map
topic/interval-transform: Galaxy genomic interval transformation pattern map
topic/sequence-transform: Galaxy sequence-record (FASTA) transformation pattern map

TDA Bioinformatics Foundry

meta Meta

Foundry-meta notes — about the Foundry's own design, machinery, or casting system rather than about topological data analysis. The one facet whose member carries no slash.

meta: Foundry-meta note — about the Foundry's own design, machinery, or casting system.

application Application not universal

The bioinformatics problem or analysis setting a note serves.

application/molecular-sciences: Molecular, biomolecular, and drug-discovery analysis.
application/single-cell: Single-cell and cellular point-cloud analysis.
application/structure-qa: Quality assessment and ranking of predicted structures and interfaces.

method Method not universal

The TDA or topological deep learning technique a note explains or implements.

method/mapper: Mapper graphs and other nerve-based topological summaries.
method/multiparameter-persistence: Persistence along more than one filtration parameter.
method/persistent-homology: Barcodes and persistence diagrams read off a filtration.
method/persistent-laplacian: Persistent topological Laplacian methods and implementations.
method/simplicial-learning: Learning directly on simplicial complexes and their Hodge structure.
method/spectral-geometry: Geometry-aware spectral and diffusion scaffolds.
method/topological-deep-learning: Learning models built on topological representations.

modality Modality not universal

The biological data modality a note consumes or describes.

modality/graph: Graphs, networks, and higher-order complexes.
modality/high-dim-tabular: High-dimensional feature matrices such as single-cell count data.
modality/molecular-structure: Three-dimensional atomic structures of molecules and complexes.
modality/point-cloud: Generic point clouds and distance matrices, without a fixed biological source.
modality/sequence: Biological sequences, as reads, proteins, or nucleotide records.

Statistical Genomics Foundry

meta Meta

Foundry-meta notes — about the Foundry's own design, tooling, or casting system rather than about statistical genomics. The one facet whose member carries no slash.

meta: Foundry-meta note — about the Foundry's own design, tooling, or casting system

domain Domain not universal

Subject-matter area a Mold/note applies to. Closed enum — add a documented row per domain.

domain/ancestral-reconstruction: Inferring ancestral sequence/character states across a phylogeny.
domain/batch-effects: Technical, non-biological variation that confounds measured signal.
domain/clustering: Unsupervised grouping into discrete categories; distance choice and cluster validation.
domain/comparative-annotation: Projecting/transferring gene annotations across genomes (liftover, CESAR, TOGA, CAT, Liftoff).
domain/data-visualization: Constructing honest exploratory/publication data displays; artifacts of scale, smoothing, ordering.
domain/differential-expression: Testing expression/abundance differences (RNA-seq, microarray, methylation).
domain/dimensionality-reduction: Ordination / low-dimensional embedding of high-dim data (PCA, SVD, MDS, t-SNE).
domain/experimental-design: Sample allocation, blocking, and randomization decided before data collection.
domain/gene-family-evolution: Gene gain/loss/duplication dynamics across lineages.
domain/image-analysis: Turning digital images into quantitative measurements: segmentation, filtering, feature extraction.
domain/molecular-evolution: Sequence change over evolutionary time (substitution, selection).
domain/pangenomics: Analysis across a species' full core/accessory gene complement.
domain/phylogenetic-comparative-methods: Trait evolution and diversification inferred on a fixed phylogeny (PGLS, BiSSE/HiSSE).
domain/phylogenetics: Inferring evolutionary trees from molecular data.
domain/single-cell-genomics: Single-cell / droplet assays (demultiplexing, hashing, count-splitting, cluster inference).
domain/spatial-statistics: Spatial point patterns vs a complete-spatial-randomness baseline (point processes, Ripley's K).
domain/statistical-inference: General validity of statistical inference — multiple testing, post-selection, confounding, reproducibility.
domain/supervised-learning: Learning a predictive rule from labeled data while guarding against overfitting.
domain/synteny: Conserved gene order / collinearity across genomes (MCScanX, SyRI, GENESPACE).
domain/whole-genome-alignment: Genome-to-genome alignment / homology mapping (Cactus, HAL, minimap2, MUMmer, chain/net).
domain/whole-genome-duplication: Polyploidy / WGD detection via Ks distributions and duplicate retention.

family Family not universal

Which side of the analyze/referee split a Mold sits on (glossary Family A / Family B).

family/a: Do the analysis (construct).
family/b: Referee the analysis (critique + calibrate).

role Role not universal

The Family-A or Family-B role a Mold plays (glossary construct/critique/calibrate).

role/construct: Family-A — frame, design-review, select, run.
role/critique: Family-B — reason about validity against known invalidity patterns.
role/calibrate: Family-B — construct and run the empirical check (null/simulation/controls/power).

topic Topic not universal

Finer-grained subject beneath a domain. Closed enum — add a documented row per topic.

topic/confounding: Batch / surrogate-variable / collider confounding of a biological contrast.
topic/dnds: Nonsynonymous/synonymous substitution-rate ratio; a test for selection (under molecular-evolution).
topic/gc-biased-gene-conversion: gBGC as a confound of selection signals.
topic/ks-distribution: Ks-based WGD dating / substitution saturation.
topic/liftover: Coordinate/annotation liftover across assemblies.
topic/multiple-testing: FWER/FDR control across many simultaneous tests.
topic/pangenome-openness: Open/closed pangenome; core/accessory dynamics.
topic/positive-selection: Codon-model tests for positive selection (BUSTED/MEME/aBSREL/branch-site).
topic/post-selection-inference: Double-dipping / selective inference / count-splitting.
topic/recombination: Recombination detection (e.g. GARD) affecting selection inference.
topic/reproducibility: Forking paths, p-hacking, researcher degrees of freedom.

Divergent facet keys: application (TDA Bioinformatics Foundry)cli (Galaxy Workflow Foundry)domain (Statistical Genomics Foundry)family (Statistical Genomics Foundry)method (TDA Bioinformatics Foundry)modality (TDA Bioinformatics Foundry)prompt (Galaxy Workflow Foundry)role (Statistical Genomics Foundry)source (Galaxy Workflow Foundry)target (Galaxy Workflow Foundry)tool (Galaxy Workflow Foundry)topic (Galaxy Workflow Foundry, Statistical Genomics Foundry)