search

monarch-initiative / SEPIO-ontology

Ontology for representing scientific evidence and provenance information
48 stars 10 forks source link

Defining ontology classes for typing of EvidenceLines #12

Open mbrush opened 7 years ago

mbrush commented 7 years ago

In adopting SEPIO, ClinGen now creates EvidenceLine objects in their model. I propose we develop a type system for classifying these objects. This will provide an important level of interoperability with other data sets with minimal E/P metadata that use only ECO codes to define a 'type' of evidnece used.

Requirements here include:

  1. Evidence classes need to be interoperable with the widespread use of evidence types from efforts like the Evidence and Conclusion Ontology (ECO). We will work with ECO as possible to extend their model to include evidence types relevant to variant pathogenicity interpretation.
  2. Evidence classes need to accommodate evidence types that are implicit or explicit in frameworks like the ACMG guidelines and Invitae's Sherloc.
  3. Evidence classes need to be consistent with the granular metadata being represented in models like ClinGen's - and created at a level that is easy to apply in practice.
mbrush commented 7 years ago

A major value of this type system will be to enable interoperability with other, less granular datasets that rely on evidence codes as a primary means of describing assertion evidence and provenance. The evidence type system should provide a framework for such data integration, as well as reasoning needed for semantic-similarity based assessment of the quality and diversity of their evidence, that will drive computational approaches to comparing and evaluating assertions.