susannasiebert
commented
6 years ago @jmcmichael pointed out that SEPIO has a really good wiki in their github repo, with a page going into some detail on how they map CIVIC evidence to their Assertions: https://github.com/monarch-initiative/SEPIO-ontology/wiki/CIViC-Variant-Interpretation
We currently use the AMP/ASCO/CAP and ACMG (Mendelian cancer predisposition) guidelines to harmonize the somatic and germline variant classification, respectively. Unfortunately, both the ACMG and AMP guidelines arguably lack the granularity to represent all of the detailed types of evidence curated in the existing cancer variant knowledgebases, and importantly do not use a formal ontology to describe this evidence. To overcome this limitation, we will adopt and extend the Monarch Initiative’s Scientific Evidence and Provenance Information Ontology (SEPIO; https://github.com/monarch-initiative/SEPIO-ontology). SEPIO was designed to represent evidence and provenance information about scientific assertions. By evolving to a SEPIO model, we will be able to incorporate more evidence models, especially those produced by machine learning, while maintaining provenance. Adopting such a model will allow us to address critical cancer variant database questions of how to weigh various evidence and handle conflicting evidence.
Additionally, to facilitate further interoperability with external CIViC data consumers—particularly clinically facing applications—we will also add support for MONDO Monarch Disease Ontology (https://www.ebi.ac.uk/ols/ontologies/mondo) , the NCI thesaurus (ncit.nci.nih.gov) and SNOMED (https://www.snomed.org/snomed-ct).