GAL system hypothesis evaluation

HyQue evaluates hypotheses using domain-specific rules for the seven event types represented in the HKB:

  1. protein-protein binding
  2. protein-nucleic acid binding
  3. molecular activation
  4. molecular inhibition
  5. gene induction
  6. gene repression
  7. transport

Each of these event types has an ideal set of criteria that provides maximum support for a hypothesis about that kind of event. This ideal set has been used to generate rule sets for HyQue. For example, the 'binds to promoter' (protein-nucleic acid binding) event has the following rule set:

  • if the event is negated (e.g. asserted as "does NOT bind to promoter) subtract 2 from score
  • if physical operator is 'promoter binding' (GO:0010843) add 1 to score
  • if actor entity is of type 'protein' (CHEBI:36080) add 1 to score; else subtract 1 from score
  • if actor entity has function 'transcription factor activity' (GO:0003702) add 1 to score
  • if location of actor entity is 'nucleus' (GO:0005634) add 1 to score; else subtract 1 from score
  • if evidence for actor entity function is 'inferred from direct assay' or 'electronic annotation' (ECO:0000002 or ECO:00000067) add 1 to score
  • if evidence for actor location is 'inferred from direct assay' (ECO:0000002) add 1 to score
  • if target entity is of type 'gene' (SO:0000236) add 1 to score; else subtract 1 from score
  • if binding site is of type 'transcription factor' (GRO:TranscriptionFactorBindingSiteOfDNA) add 1 to score; else subtract 1 from score
  • if event location is 'nucleus' (GO:0005634) add 1 to score; else subtract 1 from score

The maximum score for this event is 9. Deviations from the 'ideal' event reduce the score, and negation has the most severe penalty if events are asserted not to be negated. If an event is asserted to be negated in the hypothesis, then negation does not affect the score. Final scores are calculated as a fraction of the maximum score e.g. a score of 7 for this event is converted to 7/9.