merge.yaml

# Sourced from: https://github.com/biolink/kgx/tree/master/examples
---
configuration:
  output_directory: output/merged
  checkpoint: false

  curie_map:
  # define non-canonical CURIE to IRI mappings (for RDF)
  node_properties:
  # define predicates that are to be treated as direct node properties (for RDF)
  predicate_mappings:
  # map non-canonical predicates to a property name (for RDF)
  # property_types:
  # define the type for non-canonical properties for RDF export

merged_graph:
  name: Monarch Graph
  source:
    monarch_ontology:
      input:
        format: obojson
        filename:
          - output/monarch/monarch.json
    alliance_gene:
      input:
        format: tsv
        filename:
          - output/alliance_gene_nodes.tsv
    alliance_gene_to_phenotype:
      input:
        format: tsv
        filename:
          - output/alliance_gene_to_phenotype_edges.tsv
    alliance_publication:
      input:
        format: tsv
        filename:
          - output/alliance_publication_nodes.tsv
    ctd_chemical_to_disease:
      input:
        format: tsv
        filename:
          - output/ctd_chemical_to_disease_edges.tsv
    flybase_publication_to_gene:
      input:
        format: tsv
        filename:
          - output/flybase_publication_to_gene_edges.tsv
    goa_go_annotation:
      input:
        format: tsv
        filename:
          - output/goa_go_annotation_edges.tsv
    hgnc_gene:
      input:
        format: tsv
        filename:
          - output/hgnc_gene_nodes.tsv
          - output/hgnc_gene_edges.tsv
    hpoa_disease_phenotype:
      input:
        format: tsv
        filename:
          - output/hpoa_disease_phenotype_nodes.tsv
          - output/hpoa_disease_phenotype_edges.tsv
    mgi_publication_to_gene:
      input:
        format: tsv
        filename:
          - output/mgi_publication_to_gene_edges.tsv
    omim_gene_to_disease:
      input:
        format: tsv
        filename:
          - output/omim_gene_to_disease_nodes.tsv
          - output/omim_gene_to_disease_edges.tsv
    panther_ref_genome_orthologs:
      input:
        format: tsv
        filename:
          - output/panther_ref_genome_orthologs_nodes.tsv
          - output/panther_ref_genome_orthologs_edges.tsv
    pombase_gene:
      input:
        format: tsv
        filename:
          - output/pombase_gene_nodes.tsv
    pombase_gene_to_phenotype:
      input:
        format: tsv
        filename:
          - output/pombase_gene_to_phenotype_edges.tsv
    reactome_chemical_to_pathway:
      input:
        format: tsv
        filename:
          - output/reactome_chemical_to_pathway_edges.tsv
    reactome_gene_to_pathway:
      input:
        format: tsv
        filename:
          - output/reactome_gene_to_pathway_edges.tsv
    rgd_publication_to_gene:
      input:
        format: tsv
        filename:
          - output/rgd_publication_to_gene_edges.tsv
    sgd_publication_to_gene:
      input:
        format: tsv
        filename:
          - output/sgd_publication_to_gene_edges.tsv
    string_protein_links:
      input:
        format: tsv
        filename:
          - output/string_protein_links_edges.tsv
    xenbase_gene:
      input:
        format: tsv
        filename:
          - output/xenbase_gene_nodes.tsv
    xenbase_gene_to_phenotype:
      input:
        format: tsv
        filename:
          - output/xenbase_gene_to_phenotype_edges.tsv
    xenbase_publication_to_gene:
      input:
        format: tsv
        filename:
          - output/xenbase_publication_to_gene_edges.tsv
    zfin_gene_to_phenotype:
      input:
        format: tsv
        filename:
          - output/zfin_gene_to_phenotype_edges.tsv
    zfin_publication_to_gene:
        input:
          format: tsv
          filename:
            - output/zfin_publication_to_gene_edges.tsv
  operations:
    - name: kgx.graph_operations.summarize_graph.generate_graph_stats
      args:
        graph_name: Monarch Graph
        filename: merged_graph_stats.yaml
        node_facet_properties:
          - provided_by
        edge_facet_properties:
          - provided_by
  destination:
    merged-kg-tsv:
      format: tsv
      compression: tar.gz
      filename: monarch-kg
    merged-kg-nt:
      format: nt
      compression: gz
      filename: monarch-kg.nt.gz
#    merged-kg-neo4j:
#      format: neo4j
#      uri: bolt://localhost:7687
#      username: neo4j
#      password: admin