schema_config.yaml

Title: BioCypher graph schema configuration file for CROssBAR v2

# ---
# "Named Things"
# ---
  

gene:
  represented_as: node
  preferred_id: ncbigene
  label_in_input: gene
  properties:    
    genes: str[]
    kegg: str[]
    ensembl: str[]
    ensembl_gene_ids: str[]

protein:
  represented_as: node
  preferred_id: uniprot
  label_in_input: protein
  properties:
    secondary_ids: str[]
    length: int
    mass: int
    protein_names: str[]
    proteome: str[]
    organism_id: int
    ec: str[]
    virus_hosts: int[] 
  
organism taxon:
  represented_as: node
  preferred_id: ncbitaxon
  label_in_input: organism
  properties:
    organism: str

protein domain:  
  represented_as: node
  preferred_id: interpro
  label_in_input: domain
  properties:    
    protein_count: int
    name: str
    type: str
    parent_list: str[]
    child_list: str[]
    pfam: str[]
    ec: str[]
    pdb: str[]
    
biological process: 
  represented_as: node
  preferred_id: go
  label_in_input: biological process
  properties:
    name: str
  
cellular component:
  represented_as: node
  preferred_id: go
  label_in_input: cellular component
  properties:
    name: str

molecular function:
  synonym_for: molecular activity
  represented_as: node
  preferred_id: go
  label_in_input: molecular function
  properties:
    name: str
  
small molecule:
  represented_as: node
  preferred_id: drugbank
  label_in_input: drug
  properties:
    drugbank_id: str
    cas_number: str
    name: str
    groups: str[]
    general_references: str[]
    atc_codes: str[]
    kegg_drug: str
    rxcui: str
    pharmgkb: str
    pdb: str
    drugcentral: str
    zinc: str
    chembl: str
    bindingdb: str
    clinicaltrials: str
    chebi: str
    pubchem: str
    inchi: str
    inchikey: str

molecular mixture:
  represented_as: node
  preferred_id: id
  label_in_input: compound
  properties:
    type: str
    full_mwt: float
    species: str
    heavy_atoms: int
    alogp: float
    inchi: str
    inchikey: str
    qed_score: float

disease:
  represented_as: node
  preferred_id: mondo
  label_in_input: disease
  properties:
    name: str
    synonyms: str[]
    umls: str
    doid: str
    mesh: str
    omim: str
    efo: str
    orphanet: str
    hp: str
    icd10cm: str
    ncit: str
    icd9: str
    meddra: str

pathway:
  represented_as: node
  preferred_id: id
  label_in_input: pathway
  properties:
    name: str
    organism: str

phenotype:
  represented_as: node
  preferred_id: hp
  label_in_input: phenotype
  properties:
    name: str
    synonyms: str[]
  
# Cell: # relevant?
  # represented_as: node
  # preferred_id: CL
  # label_in_input: Cell

# cell line:
  # represented_as: node
  # preferred_id: ccle
  # label_in_input: Cell_Line

# tissue:
  # is_a: grossanatomicalstructure
  # represented_as: node
  # preferred_id: uberon
  # label_in_input: Tissue

# patient:
  # is_a: case
  # represented_as: node
  # preferred_id: id
  # label_in_input: Patient
  
# side effect:
  # is_a: phenotypic feature
  # represented_as: node
  # preferred_id: sider.effect
  # label_in_input: Side_Effect

# ---
# Associations
# ---

gene to protein relationship:
  #is_a: gene to gene product relationship
  represented_as: edge
  label_as_edge: Encodes
  preferred_id: id
  source: gene
  target: protein
  label_in_input: Encodes
  
protein to organism taxon association:
  #is_a: association
  represented_as: edge
  label_as_edge: Belongs_To
  preferred_id: id
  source: protein
  target: organismTaxon
  label_in_input: Belongs_To  
  
post translational interaction:
  #is_a: pairwise molecular interaction
  represented_as: edge
  label_as_edge: Interacts_With
  preferred_id: id
  source: protein
  target: protein
  label_in_input: Interacts_With
  properties:    
    source: str[]
    pubmed_id: int[]
    method: str
    interaction_type: str
    intact_score: float
    string_combined_score: int
    string_physical_combined_score: int

protein to protein domain association:
  #is_a: association
  represented_as: edge
  label_as_edge: protein_has_domain
  preferred_id: id
  source: protein
  target: protein domain
  label_in_input: protein_has_domain
  properties:
    locations: str[]

protein to biological process association:
  #is_a: macromolecular machine to biological process association
  represented_as: edge
  label_as_edge: protein_involved_in_biological_process
  preferred_id: go
  source: protein
  target: biological process
  label_in_input: protein_involved_in_biological_process
  properties:
    reference: str
    evidence_code: str

protein to cellular component association:
  #is_a: macromolecular machine to cellular component association
  represented_as: edge
  label_as_edge: [located_in, part_of, is_active_in]
  preferred_id: go
  source: protein
  target: cellular component
  label_in_input: [protein_located_in_cellular_component, protein_part_of_cellular_component, protein_is_active_in_cellular_component] 
  properties:
    reference: str
    evidence_code: str
  
protein to molecular function association:
  #is_a: macromolecular machine to molecular function association
  represented_as: edge
  label_as_edge: [enables, contributes_to]
  preferred_id: go
  source: protein
  target: molecular function
  label_in_input: [protein_enables_molecular_function, protein_contributes_to_molecular_function]
  properties:
    reference: str
    evidence_code: str

biological process to biological process association:
  represented_as: edge
  label_as_edge: [is_a, part_of, positively_regulates, negatively_regulates]
  preferred_id: go
  source: biological process
  target: biological process
  label_in_input: [biological_process_is_a_biological_process, biological_process_part_of_biological_process, biological_process_positively_regulates_biological_process, biological_process_negatively_regulates_biological_process]

biological process to molecular function association:
  represented_as: edge
  label_as_edge: [positively_regulates, negatively_regulates]
  preferred_id: go
  source: biological process
  target: molecular function
  label_in_input: [biological_process_positively_regulates_molecular_function, biological_process_negatively_regulates_molecular_function]

molecular function to molecular function association:
  represented_as: edge
  label_as_edge: [is_a, part_of, positively_regulates, negatively_regulates]
  preferred_id: go
  source: molecular function
  target: molecular function
  label_in_input: [molecular_function_is_a_molecular_function, molecular_function_part_of_molecular_function, molecular_function_positively_regulates_molecular_function, molecular_function_negatively_regulates_molecular_function]

cellular component to cellular component association:
  represented_as: edge
  label_as_edge: [is_a, part_of]
  preferred_id: go
  source: cellular component
  target: cellular component
  label_in_input: [cellular_component_is_a_cellular_component, cellular_component_part_of_cellular_component]

protein domain to molecular function association:
  is_a: association
  represented_as: edge
  label_as_edge: protein_domain_enables_molecular_function
  preferred_id: id
  source: protein domain
  target: molecular function
  label_in_input: protein_domain_enables_molecular_function
  
protein domain to biological process association:
  #is_a: association
  represented_as: edge
  label_as_edge: protein_domain_involved_in_biological_process
  preferred_id: id
  source: protein domain
  target: biological process
  label_in_input: protein_domain_involved_in_biological_process
  
protein domain to cellular component association:
  #is_a: association
  represented_as: edge
  label_as_edge: protein_domain_located_in_cellular_component
  preferred_id: id
  source: protein domain
  target: cellular component
  label_in_input: protein_domain_located_in_cellular_component

small molecule to protein association:
  #is_a: pairwise molecular interaction
  represented_as: edge
  label_as_edge: drug_targets_protein
  preferred_id: id
  source: small molecule
  target: protein
  label_in_input: drug_targets_protein
  properties:
    known_action: str
    activity_value: float
    confidence_score: float
    direct_interaction: bool
    disease_efficacy: bool
    mechanism_of_action: str
    pchembl: float
    activity_type: str
    dgidb_score: float
    references: str[]
    mechanism_of_action_type: str
    stitch_combined_score: float
    source: str[]

small molecule to protein association:
  #is_a: pairwise molecular interaction
  represented_as: edge
  label_as_edge: drug_targets_protein
  preferred_id: id
  source: small molecule
  target: protein
  label_in_input: drug_targets_protein
  properties:
    known_action: str
    references: str[]
    actions: str[]

small molecule to small molecule association:
  #is_a: pairwise molecular interaction
  represented_as: edge
  label_as_edge: drug_interacts_with_drug
  preferred_id: id
  source: small molecule
  target: small molecule
  label_in_input: drug_interacts_with_drug
  properties:
    recommendation: str
    interaction_level: str
    interaction_type: str[]
    source: str[]


small molecule to gene association:
  represented_as: edge
  label_as_edge: [decreases_expression, increases_expression]
  preferred_id: id
  source: small molecule
  target: gene
  label_in_input: [drug_decreases_expression_gene, drug_increases_expression_gene]
  properties:
    source: str[]
    references: str[]


molecular mixture to protein association:
  #is_a: pairwise molecular interaction
  represented_as: edge
  label_as_edge: targets
  preferred_id: id
  source: molecular mixture
  target: protein
  label_in_input: compound_targets_protein
  properties:
    source: str[]
    pchembl: float
    activity_value: float
    activity_type: str
    assay_chembl: str[]
    confidence_score: float
    stitch_combined_score: float


gene to gene orthology association:
  #is_a: association
  represented_as: edge
  label_as_edge: Is_Orthologous_With
  preferred_id: id
  source: gene
  target: gene
  label_in_input: gene_is_orthologous_with_gene
  properties:
    source: str[]
    relation_type: str
    oma_orthology_score: float

organism taxon to disease association:
  #is_a: association
  represented_as: edge
  label_as_edge: organism_causes_disease
  preferred_id: id
  source: organism taxon
  target: disease
  label_in_input: organism_causes_disease


disease to small molecule association:
  #is_a: chemical or drug or treatment to disease or phenotypic feature association
  represented_as: edge
  label_as_edge: disease_is_treated_by_drug
  preferred_id: id
  source: disease
  target: small molecule
  label_in_input: disease_is_treated_by_drug
  properties:
    source: str[]
    pubmed_ids: str[]
    max_phase: str

disease to disease hierarchical association:
  #is_a: association
  represented_as: edge
  label_as_edge: disease_is_a_disease
  preferred_id: id
  source: disease
  target: disease
  label_in_input: disease_is_a_disease

disease to disease association:
  #is_a: association
  represented_as: edge
  label_as_edge: disease_is_associated_with_disease
  preferred_id: id
  source: disease
  target: disease
  label_in_input: disease_is_associated_with_disease
  properties:
    source: str[]
    disgenet_jaccard_genes_score: float
    disgenet_jaccard_variants_score: float


disease to disease comorbidity association:
  #is_a: association
  represented_as: edge
  label_as_edge: disease_is_comorbid_with_disease
  preferred_id: id
  source: disease
  target: disease
  label_in_input: disease_is_comorbid_with_disease


gene to disease association:
  #is_a: association
  represented_as: edge
  label_as_edge: gene_is_related_to_disease
  preferred_id: id
  source: gene
  target: disease
  label_in_input: gene_is_related_to_disease
  properties:
    source: str[]
    variant_source: str[]
    opentargets_score: float
    disgenet_gene_disease_score: float
    disgenet_variant_disease_score: float
    diseases_confidence_score: float
    allele_id: str
    clinical_significance: str
    review_status: int
    variation_id: str
    pubmed_ids: str[]
    dbsnp_id: str[]


protein to pathway association:
  #is_a: association
  represented_as: edge
  label_as_edge: protein_take_part_in_pathway
  preferred_id: id
  source: protein
  target: pathway
  label_in_input: protein_take_part_in_pathway
  properties:
    source: str

disease to pathway association:
  #is_a: association
  represented_as: edge
  label_as_edge: disease_modulates_pathway
  preferred_id: id
  source: disease
  target: pathway
  label_in_input: disease_modulates_pathway
  properties:
    source: str[]

drug to pathway association:
  #is_a: association
  represented_as: edge
  label_as_edge: drug_has_target_in_pathway
  preferred_id: id
  source: drug
  target: pathway
  label_in_input: drug_has_target_in_pathway
  properties:
    source: str


pathway to pathway association:
  #is_a: association
  represented_as: edge
  label_as_edge: [pathway_is_part_of_pathway, pathway_is_equivalent_to_pathway]
  preferred_id: id
  source: pathway
  target: pathway
  label_in_input: [pathway_is_part_of_pathway, pathway_is_equivalent_to_pathway]

reactome hierarchical association:
  #is_a: association
  represented_as: edge
  label_as_edge: pathway_participates_pathway
  preferred_id: id
  source: pathway
  target: pathway
  label_in_input: pathway_participates_pathway

protein to phenotype association:
  #is_a: association
  represented_as: edge
  label_as_edge: protein_is_associated_with_phenotype
  preferred_id: id
  source: protein
  target: phenotype
  label_in_input: protein_is_associated_with_phenotype

phenotype hierarchical association:
  #is_a: association
  represented_as: edge
  label_as_edge: phenotype_is_a_phenotype
  preferred_id: id
  source: phenotype
  target: phenotype
  label_in_input: phenotype_is_a_phenotype

phenotype to disease association:
  #is_a: association
  represented_as: edge
  label_as_edge: phenotype_is_associated_with_disease
  preferred_id: id
  source: phenotype
  target: disease
  label_in_input: phenotype_is_associated_with_disease
  properties:
    pubmed_ids: str[]
    evidence: str

# protein to protein domain association:
  # represented_as: edge
  # label_as_edge: Has_Domain
  # preferred_id: id
  # source: protein
  # target: protein domain
  # label_in_input: Protein_Has_Domain

# protein to pathway association:
  # represented_as: edge
  # label_as_edge: Takes_Part_In
  # preferred_id: id
  # source: protein
  # target: pathway
  # label_in_input: Take_Part_In

# protein to disease association:
  # represented_as: edge
  # label_as_edge: Is_Targeted_In
  # preferred_id: id
  # source: protein
  # target: disease
  # label_in_input: Protein_Is_Related_To_Disease

# gene to gene homology association:
  # represented_as: edge
  # label_as_edge: Is_Ortholog_To
  # preferred_id: id
  # source: gene
  # target: gene
  # label_in_input: Is_Ortholog_To

# pairwise gene to gene interaction:
  # represented_as: edge
  # label_as_edge: Regulates
  # preferred_id: id
  # source: gene
  # target: gene
  # label_in_input: Regulates

# mutation to tissue association:
  # is_a: [genotype to tissue association, association]
  # represented_as: edge
  # label_as_edge: Is_Mutated_In
  # preferred_id: id
  # source: gene
  # target: tissue
  # label_in_input: Gene_Is_Mutated_In_Tissue

# mutation to cell line association:
  # is_a: [genotype to cell line association, association]
  # represented_as: edge
  # label_as_edge: Is_Mutated_In
  # preferred_id: id
  # source: gene
  # target: cell line
  # label_in_input: Gene_Is_Mutated_In_Cell_Line

# expression to tissue association:
  # is_a: [genotype to tissue association, association]
  # represented_as: edge
  # label_as_edge: Is_DEG_In
  # preferred_id: id
  # source: gene
  # target: tissue
  # label_in_input: Gene_Is_DEG_In_Tissue

# expression to cell line association:
  # is_a: [genotype to cell line association, association]
  # represented_as: edge
  # label_as_edge: Is_DEG_In
  # preferred_id: id
  # source: gene
  # target: cell line
# label_in_input: Gene_Is_DEG_In_Cell_Line

# gene to phenotypic feature association:
  # represented_as: edge
  # label_as_edge: Is_Associated_With
  # preferred_id: id
  # source: gene
  # target: phenotypic feature
  # label_in_input: Gene_Is_Associated_With_Phenotype

# mutation to case association:
  # is_a: association
  # represented_as: edge
  # label_as_edge: Is_Mutated_In
  # preferred_id: id
  # source: gene
  # target: case
  # label_in_input: Gene_Is_Mutated_In_Patient

# expression to case association:
  # is_a: association
  # represented_as: edge
  # label_as_edge: Is_DEG_In
  # preferred_id: id
  # source: gene
  # target: case
  # label_in_input: Gene_Is_DEG_In_Patient

# gene to pathway association:
  # represented_as: edge
  # label_as_edge: Is_Member_Of
  # source: protein
  # target: pathway
  # preferred_id: id
  # label_in_input: Is_Member_Of

# gene to disease association:
  # represented_as: edge
  # label_as_edge: Is_Related_To
  # preferred_id: id
  # source: gene
  # target: disease
  # label_in_input: Gene_Is_Related_To_Disease

# drug interaction:
  # is_a: chemical to chemical association
  # represented_as: edge
  # label_as_edge: Interacts_With
  # preferred_id: id
  # source: small molecule
  # target: small molecule
  # label_in_input: Drug_Interacts_With_Drug

# small molecule to side effect association:
  # represented_as: edge
  # label_as_edge: Has_Side_Effect
  # preferred_id: id
  # source: small molecule
  # target: side effect
  # label_in_input: Drug_Has_Side_Effect

# small molecule to tissue association:
  # represented_as: edge
  # preferred_id: id
  # source: small molecule
  # target: tissue
  # label_in_input: Drug_Targets_Tissue

# small molecule to cell line association:
  # represented_as: edge
  # preferred_id: id
  # source: small molecule
  # target: tissue
  # label_in_input: Drug_Targets_Cell_Line

# small molecule to pathway association:
  # represented_as: edge
  # label_as_edge: Has_Target_In
  # preferred_id: id
  # source: small molecule
  # target: pathway
  # label_in_input: Has_Target_In

# molecular mixture to protein association:
  # represented_as: edge
  # label_as_edge: Compound_Targets_Protein
  # preferred_id: id
  # source: molecular mixture
  # target: protein
  # label_in_input: Compound_Targets_Protein

# tissue to disease association:
  # represented_as: edge
  # label_as_edge: Has_Disease
  # preferred_id: id
  # source: tissue
  # target: disease
  # label_in_input: Tissue_Has_Disease

# cell line as a model of disease association:
  # represented_as: edge
  # preferred_id: id
  # source: cell line
  # target: disease
  # label_in_input: Cell_Line_Has_Disease

# case to disease association:
  # represented_as: edge
  # label_as_edge: Has_Disease
  # preferred_id: id
  # source: patient
  # target: disease
  # label_in_input: Patient_Has_Disease

# disease to disease association:
  # represented_as: edge
  # label_as_edge: Comorbid_With
  # preferred_id: id
  # source: disease
  # target: disease
  # label_in_input: Comorbid_With

# disease to phenotypic feature association:
  # represented_as: edge
  # label_as_edge: Is_Associated_With
  # preferred_id: id
  # source: disease
  # target: phenotypic feature
  # label_in_input: Disease_Is_Associated_With_Phenotype

# disease to pathway association:
  # represented_as: edge
  # label_as_edge: Modulates
  # preferred_id: id
  # source: disease
  # target: pathway
  # label_in_input: Modulates

# disease to small molecule association:
  # represented_as: edge
  # label_as_edge: Is_Treated_By
  # preferred_id: id
  # source: disease
  # target: small molecule
  # label_in_input: Is_Treated_By

# protein domain to molecular activity association:
  # represented_as: edge
  # label_as_edge: Has_Function
  # preferred_id: id
  # source: protein domain
  # target: molecular activity
  # label_in_input: Domain_Has_Function

# protein domain to cellular component association:
  # represented_as: edge
  # preferred_id: id
  # source: protein domain
  # target: cellular component
  # label_in_input: Domain_Has_Location

# suggestion for drug - small molecule
# drug to small molecule association:
  # is_a: association
  # represented_as: edge
  # label_as_edge: Contains_Active_Ingredient
  # preferred_id: id
  # source: drug
  # target: small molecule
  # label_in_input: Contains_Active_Ingredient