Pattern analysis

This repository contains the ongoing work of extracting and documenting phenotype related patterns across relevant phenotype ontologies

Our toolkit aims at supporting the following tasks:

• We call
  • CORPUS a set of ontologies
  • O an ontology
  • AX an axiom
  • B a branch of an ontology, i.e. the set of indirect subclasses of a particular class
  • MAPPING a mapping between sets of terms
• Harmonising ontology patterns
  • within ontologies
  • across ontologies
  • by branch (default: owl:Thing)
  • by
    • Identifying patterns
      1. DEFINITIONS = CORPUS.O.AX.filter(AX of the form A = C, where C is a complex class expression ): identifying all equivalent class axioms across all O in CORPUS.
      2. GENERATED = For each definition D in DEFINITIONS, instantiate a set of more general patterns by gradually replacing all class names in D with a more general one (superclass in O)
    • Quantifying their impact
      1. IMPACT1: For each pattern P in GENERATED, count the number of classes in B (branch) that are * direct subclasses of P * indirect subclasses of P
    • Presenting the patterns and impact in an easy to navigate way
      • By exporting the patterns as an OWL ontology
        • For convenience, we also export the inferred class hierarchy
      • Given a fast enough reasoner, such as Fact++ the patterns can be queried using the DL query tab by
        • Asking for all patterns with a particular impact, i.e. hasImpact some integer[>100].
        • For convenience, we add a few defined classes with some default values for these.
    • Suggesting patterns to be replaced:
      • Based on MAPPING,