AI Drug Discovery Ontology

AI SEARCH

FLOW ALGEBRA — AI DRUG DISCOVERY ONTOLOGY
Semantic Library
Version: 4.0
Canonical Page:
https://kirethresearch.org/booksai-core/ontologies/biological-ontologies/ai-biology/ai-drug-discovery-ontology.html

FLOW AI_DRUG_DISCOVERY_ONTOLOGY {

  META {
    DOMAIN = "Biological Ontologies"
    SUBDOMAIN = "AI Biology"
    ONTOLOGY = "AI Drug Discovery Ontology"
    PURPOSE = "Formalise entities, relationships, computational processes and reasoning structures connecting artificial intelligence with drug discovery, molecular biology, pharmacology and therapeutic development."
    LOCALE = "en-GB"
  }

  ENTITY_TYPES {
    AI_SYSTEM
    AI_MODEL
    MACHINE_LEARNING_MODEL
    DEEP_LEARNING_MODEL
    FOUNDATION_MODEL
    MOLECULAR_MODEL
    DRUG
    DRUG_CANDIDATE
    COMPOUND
    SMALL_MOLECULE
    BIOLOGIC
    MOLECULE
    CHEMICAL_STRUCTURE
    MOLECULAR_FEATURE
    MOLECULAR_DESCRIPTOR
    MOLECULAR_EMBEDDING
    TARGET
    DRUG_TARGET
    PROTEIN
    GENE
    RECEPTOR
    ENZYME
    PATHWAY
    CELL
    TISSUE
    ORGANISM
    DISEASE
    PHENOTYPE
    BIOMARKER
    ASSAY
    SCREEN
    DATASET
    KNOWLEDGE_GRAPH
    BIOLOGICAL_ONTOLOGY
    BINDING_SITE
    LIGAND
    INTERACTION
    BINDING_AFFINITY
    ACTIVITY
    EFFICACY
    POTENCY
    SELECTIVITY
    TOXICITY
    SAFETY_SIGNAL
    ADME_PROPERTY
    PHARMACOKINETIC_PROPERTY
    PHARMACODYNAMIC_PROPERTY
    FORMULATION
    DOSE
    RESPONSE
    PREDICTION
    CLASSIFICATION
    RANKING
    OPTIMISATION_RESULT
    EVIDENCE
    CONFIDENCE_SCORE
    PROVENANCE_RECORD
  }

  RELATIONSHIPS {
    TARGETS
    BINDS_TO
    INTERACTS_WITH
    INHIBITS
    ACTIVATES
    MODULATES
    REGULATES
    ENCODES
    PARTICIPATES_IN
    ASSOCIATED_WITH
    TREATS
    INVESTIGATED_FOR
    HAS_STRUCTURE
    HAS_FEATURE
    HAS_DESCRIPTOR
    HAS_EMBEDDING
    HAS_ACTIVITY
    HAS_AFFINITY
    HAS_POTENCY
    HAS_SELECTIVITY
    HAS_TOXICITY
    HAS_ADME_PROPERTY
    HAS_PHARMACOKINETIC_PROPERTY
    HAS_PHARMACODYNAMIC_PROPERTY
    PREDICTS
    CLASSIFIES
    RANKS
    GENERATES
    OPTIMISES
    PRIORITISES
    DERIVED_FROM
    TRAINED_ON
    VALIDATED_ON
    INFERRED_FROM
    SUPPORTED_BY
    CONTRADICTED_BY
    HAS_CONFIDENCE
    HAS_PROVENANCE
    MAPS_TO
    SEMANTICALLY_EQUIVALENT_TO
  }

  INPUTS {
    CHEMICAL_STRUCTURES
    MOLECULAR_DESCRIPTORS
    MOLECULAR_GRAPHS
    PROTEIN_SEQUENCES
    PROTEIN_STRUCTURES
    TARGET_DATA
    GENOMIC_DATA
    TRANSCRIPTOMIC_DATA
    PROTEOMIC_DATA
    PHENOTYPE_DATA
    DISEASE_DATA
    ASSAY_DATA
    SCREENING_DATA
    BIOACTIVITY_DATA
    TOXICITY_DATA
    ADME_DATA
    PHARMACOKINETIC_DATA
    PHARMACODYNAMIC_DATA
    CLINICAL_DATA
    LITERATURE_EVIDENCE
    BIOLOGICAL_KNOWLEDGE_GRAPHS
    ONTOLOGY_TERMS
  }

  NORMALISE {
    STANDARDISE_IDENTIFIERS
    NORMALISE_CHEMICAL_STRUCTURES
    NORMALISE_TARGET_IDENTIFIERS
    NORMALISE_ASSAY_UNITS
    NORMALISE_ACTIVITY_MEASURES
    MAP_DISEASE_TERMS
    MAP_GENE_AND_PROTEIN_IDENTIFIERS
    ALIGN_ONTOLOGY_TERMS
    REMOVE_DUPLICATES
    VALIDATE_METADATA
    PRESERVE_PROVENANCE
  }

  PROCESS DRUG_DISCOVERY_PIPELINE {
    INPUT -> DATA_QUALITY_CONTROL
    DATA_QUALITY_CONTROL -> PREPROCESSING
    PREPROCESSING -> FEATURE_EXTRACTION
    FEATURE_EXTRACTION -> REPRESENTATION_LEARNING
    REPRESENTATION_LEARNING -> MODEL_INFERENCE
    MODEL_INFERENCE -> CANDIDATE_GENERATION
    CANDIDATE_GENERATION -> CANDIDATE_PRIORITISATION
    CANDIDATE_PRIORITISATION -> BIOLOGICAL_INTERPRETATION
    BIOLOGICAL_INTERPRETATION -> EVIDENCE_INTEGRATION
    EVIDENCE_INTEGRATION -> CONFIDENCE_ESTIMATION
    CONFIDENCE_ESTIMATION -> OUTPUT
  }

  PROCESS TARGET_DISCOVERY {
    DISEASE_DATA + GENOMIC_DATA + PROTEOMIC_DATA + PATHWAY_DATA
      -> TARGET_IDENTIFICATION

    TARGET_IDENTIFICATION
      -> TARGET_VALIDATION

    TARGET_VALIDATION
      -> TARGET_PRIORITISATION

    TARGET_PRIORITISATION
      -> DRUGGABILITY_ASSESSMENT
      -> CONFIDENCE_SCORE
  }

  PROCESS VIRTUAL_SCREENING {
    COMPOUND_LIBRARY + DRUG_TARGET
      -> MOLECULAR_REPRESENTATION

    MOLECULAR_REPRESENTATION
      -> INTERACTION_PREDICTION

    INTERACTION_PREDICTION
      -> BINDING_AFFINITY_PREDICTION

    BINDING_AFFINITY_PREDICTION
      -> CANDIDATE_RANKING
  }

  PROCESS DE_NOVO_DRUG_DESIGN {
    TARGET + DESIGN_CONSTRAINTS
      -> GENERATIVE_MODEL

    GENERATIVE_MODEL
      -> GENERATED_COMPOUNDS

    GENERATED_COMPOUNDS
      -> PROPERTY_PREDICTION
      -> SYNTHETIC_FEASIBILITY
      -> SAFETY_PREDICTION
      -> ACTIVITY_PREDICTION
      -> MULTI_OBJECTIVE_OPTIMISATION

    MULTI_OBJECTIVE_OPTIMISATION
      -> PRIORITISED_DRUG_CANDIDATES
  }

  PROCESS ADMET_PREDICTION {
    DRUG_CANDIDATE
      -> ABSORPTION_PREDICTION
      -> DISTRIBUTION_PREDICTION
      -> METABOLISM_PREDICTION
      -> EXCRETION_PREDICTION
      -> TOXICITY_PREDICTION

    ABSORPTION_PREDICTION
    + DISTRIBUTION_PREDICTION
    + METABOLISM_PREDICTION
    + EXCRETION_PREDICTION
    + TOXICITY_PREDICTION
      -> ADMET_PROFILE
  }

  PROCESS DRUG_REPURPOSING {
    EXISTING_DRUG + DISEASE + TARGET + KNOWLEDGE_GRAPH
      -> RELATION_INFERENCE

    RELATION_INFERENCE
      -> REPURPOSING_HYPOTHESIS

    REPURPOSING_HYPOTHESIS
      -> EVIDENCE_INTEGRATION
      -> CANDIDATE_PRIORITISATION
  }

  PROCESS KNOWLEDGE_GRAPH_REASONING {
    DRUG + TARGET + GENE + PROTEIN + PATHWAY + DISEASE + EVIDENCE
      -> DRUG_DISCOVERY_KNOWLEDGE_GRAPH

    DRUG_DISCOVERY_KNOWLEDGE_GRAPH
      -> ENTITY_RESOLUTION
      -> LINK_PREDICTION
      -> RELATION_INFERENCE
      -> EXPLANATION_PATH
  }

  AI_TASKS {
    TARGET_IDENTIFICATION
    TARGET_PRIORITISATION
    DRUG_TARGET_INTERACTION_PREDICTION
    BINDING_AFFINITY_PREDICTION
    VIRTUAL_SCREENING
    MOLECULAR_PROPERTY_PREDICTION
    BIOACTIVITY_PREDICTION
    TOXICITY_PREDICTION
    ADMET_PREDICTION
    DE_NOVO_MOLECULE_GENERATION
    LEAD_OPTIMISATION
    DRUG_REPURPOSING
    BIOMARKER_DISCOVERY
    PATIENT_STRATIFICATION
    RESPONSE_PREDICTION
    KNOWLEDGE_GRAPH_COMPLETION
    LITERATURE_MINING
  }

  RULES {
    IF ENTITY == DRUG_CANDIDATE
    THEN REQUIRE {
      CHEMICAL_OR_BIOLOGICAL_IDENTITY
      TARGET_OR_PHENOTYPIC_CONTEXT
      EVIDENCE
      PROVENANCE
    }

    IF AI_MODEL PREDICTS DRUG_TARGET_INTERACTION
    THEN LINK {
      DRUG_CANDIDATE
      TARGET
      PREDICTED_INTERACTION
      CONFIDENCE_SCORE
      SUPPORTING_EVIDENCE
    }

    IF MODEL_OUTPUT == PREDICTION
    THEN DO_NOT_ASSERT_AS_EXPERIMENTALLY_VALIDATED_FACT

    IF RELATION == TREATS
    THEN REQUIRE {
      APPROPRIATE_EVIDENCE
      INDICATION_CONTEXT
      PROVENANCE_RECORD
    }

    IF COMPOUND_HAS_HIGH_PREDICTED_ACTIVITY
    THEN DO_NOT_INFER_ACCEPTABLE_SAFETY

    IF IN_VITRO_ACTIVITY_EXISTS
    THEN DO_NOT_INFER_IN_VIVO_EFFICACY

    IF PRECLINICAL_EVIDENCE_EXISTS
    THEN DO_NOT_INFER_CLINICAL_EFFICACY

    IF SEMANTIC_IDENTIFIERS_DIFFER
    AND REFERENT_IS_EQUIVALENT
    THEN APPLY SEMANTIC_EQUIVALENCE
  }

  CONFIDENCE {
    CONFIDENCE_SCORE =
      WEIGHT(
        DATA_QUALITY,
        MODEL_VALIDATION,
        EXPERIMENTAL_SUPPORT,
        INDEPENDENT_REPLICATION,
        BIOLOGICAL_PLAUSIBILITY,
        ASSAY_RELIABILITY,
        DATASET_REPRESENTATION,
        PROVENANCE_COMPLETENESS
      )

    LEVELS {
      HIGH
      MODERATE
      LOW
      INSUFFICIENT
    }

    IF CONFLICTING_EVIDENCE
    THEN REDUCE CONFIDENCE_SCORE

    IF ORTHOGONAL_EXPERIMENTAL_VALIDATION
    THEN INCREASE CONFIDENCE_SCORE

    IF INDEPENDENT_REPLICATION
    THEN INCREASE CONFIDENCE_SCORE
  }

  SEMANTIC_EQUIVALENCE {
    EQUIVALENCE_KEY =
      NORMALISE(
        ENTITY_TYPE,
        CANONICAL_IDENTIFIER,
        CHEMICAL_STRUCTURE,
        BIOLOGICAL_TARGET,
        DISEASE_CONTEXT
      )

    IF EQUIVALENCE_KEY(A) == EQUIVALENCE_KEY(B)
    THEN A SEMANTICALLY_EQUIVALENT_TO B

    ELSE PRESERVE_DISTINCT_ENTITIES
  }

  REASONING_CONSTRAINTS {
    PREDICTION != EXPERIMENTAL_OBSERVATION
    ASSOCIATION != CAUSATION
    BINDING != THERAPEUTIC_EFFICACY
    POTENCY != SAFETY
    IN_VITRO_RESULT != CLINICAL_OUTCOME
    PRECLINICAL_RESULT != CLINICAL_VALIDATION
    MODEL_CONFIDENCE != BIOLOGICAL_CERTAINTY
    DRUG_EFFECT_IS_CONTEXT_DEPENDENT
    DOSE_AND_EXPOSURE_MATTER
    PROVENANCE_MUST_BE_PRESERVED
    UNCERTAINTY_MUST_NOT_BE_DISCARDED
  }

  VALIDATION {
    CHECK_ENTITY_TYPES
    CHECK_RELATION_DOMAINS_AND_RANGES
    CHECK_CHEMICAL_IDENTIFIERS
    CHECK_TARGET_IDENTIFIERS
    CHECK_ASSAY_CONTEXT
    CHECK_EVIDENCE_LINKS
    CHECK_PROVENANCE
    CHECK_CONFIDENCE
    CHECK_SEMANTIC_CONSISTENCY
    CHECK_DUPLICATE_ENTITIES
    CHECK_UNSUPPORTED_THERAPEUTIC_ASSERTIONS
  }

  OUTPUTS {
    NORMALISED_DRUG_ENTITIES
    NORMALISED_TARGET_ENTITIES
    DRUG_TARGET_RELATIONSHIPS
    TARGET_PRIORITIES
    CANDIDATE_DRUGS
    BINDING_PREDICTIONS
    ACTIVITY_PREDICTIONS
    ADMET_PREDICTIONS
    TOXICITY_PREDICTIONS
    REPURPOSING_HYPOTHESES
    KNOWLEDGE_GRAPH_EDGES
    CONFIDENCE_SCORES
    EVIDENCE_LINKS
    PROVENANCE_RECORDS
    EXPLANATION_PATHS
  }

  UPDATE_MODEL {
    ON NEW_EXPERIMENTAL_DATA {
      NORMALISE
      ENTITY_RESOLUTION
      UPDATE_FEATURES
      UPDATE_RELATIONSHIPS
      UPDATE_CONFIDENCE
      PRESERVE_VERSION_HISTORY
    }

    ON NEW_EVIDENCE {
      LINK_EVIDENCE
      REASSESS_SUPPORT
      REASSESS_CONTRADICTIONS
      UPDATE_CONFIDENCE
      RECORD_PROVENANCE
    }

    ON ONTOLOGY_REVISION {
      REMAP_TERMS
      CHECK_SEMANTIC_EQUIVALENCE
      IDENTIFY_DEPRECATED_TERMS
      MIGRATE_RELATIONSHIPS
      PRESERVE_PRIOR_VERSION_REFERENCES
    }
  }

  FLOW_HASH {
    HASH_INPUT =
      CANONICALISE(
        ENTITY_TYPES,
        RELATIONSHIPS,
        RULES,
        CONFIDENCE,
        SEMANTIC_EQUIVALENCE,
        REASONING_CONSTRAINTS,
        UPDATE_MODEL
      )

    FLOW_HASH = SHA256(HASH_INPUT)
  }

  AI_CONSUMPTION {
    INDEX {
      ONTOLOGY
      ENTITY_TYPES
      RELATIONSHIPS
      AI_TASKS
      RULES
      OUTPUTS
    }

    RETRIEVE {
      BY_DRUG
      BY_COMPOUND
      BY_TARGET
      BY_DISEASE
      BY_RELATION
      BY_AI_TASK
      BY_EVIDENCE
      BY_CONFIDENCE
    }

    REASON {
      APPLY_RELATIONSHIPS
      APPLY_RULES
      PRESERVE_UNCERTAINTY
      TRACE_EVIDENCE
      TRACE_PROVENANCE
      GENERATE_EXPLANATION_PATH
    }

    RESPONSE_POLICY {
      DISTINGUISH_PREDICTED_FROM_OBSERVED
      DISTINGUISH_PRECLINICAL_FROM_CLINICAL
      RETURN_CONFIDENCE_WHEN_AVAILABLE
      RETURN_PROVENANCE_WHEN_AVAILABLE
      EXPOSE_CONFLICTING_EVIDENCE
      AVOID_UNSUPPORTED_THERAPEUTIC_INFERENCE
    }
  }

  TERMINAL {
    RESULT =
      VALIDATED(
        SEMANTICALLY_NORMALISED(
          EVIDENCE_LINKED(
            CONFIDENCE_AWARE(
              AI_DRUG_DISCOVERY_KNOWLEDGE
            )
          )
        )
      )
  }
}

END FLOW