Skip to content

alliance-ingest

The Alliance of Genome Resources contains a subset of model organism data from member databases that is harmonized to the same model. The Alliance has bulk data downloads, ingest data formats, and an API. In some cases it may continue to be more practical to load from individual MODs when data is not yet fully harmonized in the Alliance.

Data Source

Data is downloaded from: https://fms.alliancegenome.org/download/

Species Included

  • Human (NCBITaxon:9606)
  • Mouse (NCBITaxon:10090)
  • Rat (NCBITaxon:10116)
  • Zebrafish (NCBITaxon:7955)
  • Fruit fly (NCBITaxon:7227)
  • C. elegans (NCBITaxon:6239)
  • Baker's yeast (NCBITaxon:4932)
  • Frog (NCBITaxon:8364)

Data sources: MGI, RGD, WormBase, FlyBase, ZFIN, SGD, Xenbase

Output

Gene

Gene entities from BGI (Basic Gene Information) ingest files for all Alliance species.

Biolink captured:

  • biolink:Gene
    • id (gene CURIE)
    • symbol
    • name
    • type (SO term for gene type)
    • in_taxon (NCBI Taxon ID)
    • in_taxon_label (species name)
    • xref (cross-references)
    • synonym
    • provided_by (source database)

Disease

Disease associations from the combined Alliance disease file. Supports gene, allele, and genotype (affected genomic model) subjects. Only "is_model_of" association types are currently processed; rows with experimental conditions (except "standard conditions") or modifiers are filtered.

Biolink captured:

  • biolink:GeneToDiseaseAssociation

    • id (UUID)
    • subject (gene ID)
    • predicate (biolink:related_to)
    • object (DOID)
    • has_evidence (evidence code)
    • publications
    • primary_knowledge_source (source database)
    • aggregator_knowledge_source (["infores:monarchinitiative", "infores:agrkb"])

  • biolink:VariantToDiseaseAssociation

    • id (UUID)
    • subject (allele ID)
    • predicate (biolink:related_to)
    • object (DOID)
    • has_evidence, publications, knowledge sources (as above)

  • biolink:GenotypeToDiseaseAssociation

    • id (UUID)
    • subject (affected genomic model ID)
    • predicate (biolink:model_of)
    • object (DOID)
    • has_evidence, publications, knowledge sources (as above)

Phenotype

Phenotype associations using the phenotype ingest format. This file contains gene, allele, and genotype phenotypes. An entity lookup file is used to determine subject category. Environmental conditions are captured as qualifiers.

RGD, MGI and WormBase records carry a single pre-composed phenotype term (MP, WBPhenotype), which is used directly as the object.

FlyBase and SGD annotate phenotypes post-composed: an ordered list of terms (FBbt/GO head term plus FBcv qualifiers; APO qualifier plus APO observable) and, for SGD, often a ChEBI chemical condition. uphenotizer pre-composes a class for each annotation shape (PHENOTYPE: prefix; the DPO-ext and APO-ext components imported into phenio) and publishes a binding table, output/source_map.tsv, keyed on (source, ordered terms, chemical). This ingest downloads that table and loads it as a koza mapping (uphenotizer_phenotype_lookup.yaml); phenotype_binding.py builds the same key from each Alliance record. A record resolves to:

  • the bound PHENOTYPE: class, for a routed binding
  • the head FBcv term itself, for a covered_upstream binding (the head term is already a DPO class, which uphenotizer leaves to DPO), but only when every trailing term is context that uphenotizer itself drops when routing: dominance, clone, sex, temporal or genotype-to-phenotype qualifiers (by FBcv namespace, read from fbcv.obo) or an FBdv stage. A trailing GO process, second phenotypic class, or spatial/intensity/extent qualifier would make the head term lossy, so those records are dropped and counted as covered_upstream_unhandled_qualifiers
  • nothing, for unrouted/skipped bindings or shapes not yet in the table (counted per source and status in the log at the end of the transform)

An SGD record with several chemical conditions yields one association per chemical. The binding table is recomputed weekly by uphenotizer, so coverage grows as rules are added there.

Biolink captured:

  • biolink:GeneToPhenotypicFeatureAssociation

    • id (UUID)
    • subject (gene ID)
    • predicate (biolink:has_phenotype)
    • object (phenotype term ID)
    • publications
    • qualifiers (condition class IDs)
    • primary_knowledge_source (source database)
    • aggregator_knowledge_source (["infores:monarchinitiative", "infores:agrkb"])

  • biolink:GenotypeToPhenotypicFeatureAssociation

    • subject (genotype ID), fields as above

  • biolink:VariantToPhenotypicFeatureAssociation

    • subject (variant ID), fields as above

Expression

Gene expression data. The full data model of the Alliance expression file includes Species, GeneID, GeneSymbol, Location, StageTerm, AssayID, CellularComponentID, AnatomyTermID, and more. Not all fields are currently populated in all input data sets.

Discussion Group: https://www.alliancegenome.org/working-groups#expression Download: https://www.alliancegenome.org/downloads#expression

Biolink captured:

  • biolink:GeneToExpressionSiteAssociation
    • id (UUID)
    • subject (gene ID)
    • predicate (biolink:expressed_in)
    • object (anatomical structure term ID or cellular component term ID)
    • stage_qualifier (stage term ID, when available)
    • qualifiers (assay type)
    • publications
    • primary_knowledge_source (source database)
    • aggregator_knowledge_source (["infores:monarchinitiative", "infores:agrkb"])

Genotype

Genotype/AGM (Affected Genomic Model) entities and their associations to alleles and genes.

Biolink captured:

  • biolink:Genotype

    • id (genotype ID)
    • name
    • type (subtype)
    • in_taxon (NCBI Taxon ID)
    • in_taxon_label

  • biolink:GenotypeToVariantAssociation

    • id (UUID)
    • subject (genotype ID)
    • predicate (biolink:has_sequence_variant)
    • object (allele ID)
    • qualifier (zygosity)
    • primary_knowledge_source, aggregator_knowledge_source

  • biolink:GenotypeToGeneAssociation

    • id (UUID)
    • subject (genotype ID)
    • predicate (biolink:related_to)
    • object (gene ID, from allele lookup)
    • primary_knowledge_source, aggregator_knowledge_source

Allele

Allele/variant entities and their associations to genes.

Biolink captured:

  • biolink:SequenceVariant

    • id (allele ID)
    • name (allele symbol)
    • in_taxon (NCBI Taxon ID)
    • in_taxon_label
    • synonym

  • biolink:VariantToGeneAssociation

    • id (UUID)
    • subject (allele ID)
    • predicate (biolink:is_sequence_variant_of)
    • original_predicate (variant type SO term)
    • object (gene ID)
    • primary_knowledge_source (source database)
    • aggregator_knowledge_source (["infores:monarchinitiative", "infores:agrkb"])

Post-Download Processing

The postdownload step extracts entity IDs for lookup files used by the phenotype and genotype transforms: - data/alliance_gene.tsv - Gene IDs - data/alliance_allele.tsv - Allele IDs - data/alliance_genotype.tsv - Genotype IDs

Usage

# Install dependencies
just install

# Run full pipeline
just run

# Or run steps individually
just download        # Download Alliance data files
just postdownload    # Extract entity lookup files
just transform-all   # Run all Koza transforms
just test            # Run tests

# Run specific transform
just transform gene
just transform disease

Requirements

  • Python 3.10+
  • uv package manager
  • just command runner

Citation

Harmonizing model organism data in the Alliance of Genome Resources. 2022. Alliance of Genome Resources Consortium. Genetics, Volume 220, Issue 4, April 2022. Published Online: 25 February 2022. doi: doi.org/10.1093/genetics/iyac022. PMID: 35380658; PMCID: PMC8982023.

License

BSD-3-Clause