-
Notifications
You must be signed in to change notification settings - Fork 11
Old CIViC Variant Interpretation
The Clinical Interpretations of Variants in Cancer (CIViC) database is a crowd-sourced database that curates evidence from the literature that support or refute cancer variant interpretations - i.e. assertions about the clinical relevance of genetic variants to cancer.
CIViC curators generate "evidence statements", which describe a study finding as a putative line of evidence for or against a particular variant interpretation. While the underlying data items that support these findings is summarized only as free text, CIVic provides some structured metadata for each evidence statement, including:
- The publication from which the evidence statement was curated
- An evidence level indicating the type and significance of the study generating evidence
- A trust rating reflecting the quality of the evidence the curated Finding provides
- The agents and dates of contributions to evidence statements across its lifecycle (as it is submitted, accepted, modified, and reviewed in the system).
The exemplar assertion modeled below is a variant interpretation that "The BRAF V600E variant correlates with sensitivity of colorectal cancer to combination therapy with Vemurafenib + Panitumumab". This variant interpretation is based on there being at least one evidence statement describing a specific finding that supports it. In the example, this is a case report describing a colorectal cancer patient with this variant who responded to this particular treatment. For the example assertion there is also an evidence statement that does not support the variant interpretation, based on a clinical study reporting the majority of subjects failed to respond to this treatment.
This example is valuable because it illustrates how SEPIO models conflicting lines of evidence for a particular claim, and assertions about the strength of support offered by a particular Evidence Line. It also highlights the distinction between statements that represent study findings and those representing evidence-based assertions.
View raw image here.
In the SEPIO model above, the main assertion is linked to two evidence lines - one supporting the proposition it puts forth and one disputing it. The asserted strength of each evidence line is captured using scores from CIViC's '5-star' system. Each CIViC evidence statement is represented as a study finding, and linked to the type of study that generated it, and the publication that described it. Finally, SEPIO can represent the various contributions made by curators to the interpretation - shown here for the curator who initially accepted the each evidence statement into the system.
Graph Description:
- The RDF data depicted above was created by the Monarch Initiative ETL pipeline.
- Each node in the graph represents an instance (i.e. an OWL individual) with its IRI (and a label where applicable) shown in bold, and its type IRI and label shown non-bolded below.
- IRIs are shown as compact CURIEs (link), whose prefix expansions are shown below.
- Instance IRIs are based on existing identifier systems where possible (e.g. IMPC, IMPRESS). Where no such identifiers exist for a node, it is treated as an anonymous individual (i.e. 'blank node'), or given a hash-based IRI in the Monarch namespace.
Prefix Expansions:
- MONARCH: http://www.monarchinitiative.org/MONARCH_
- RO: http://purl.obolibrary.org/obo/RO_
- PMID: http://www.ncbi.nlm.nih.gov/pubmed/
- FOAF: http://xmlns.com/foaf/0.1/
- CIVIC-variant: http://civic.genome.wustl.edu/links/variants/
- CIVIC-evidence: http://civic.genome.wustl.edu/links/evidence/
- CIVIC-drug: http://civic.genome.wustl.edu/links/drug/
- CIVIC-agent: http://civic.genome.wustl.edu/links/agent/
A Note on Assertion Semantics:
The meaning of the Assertion in the graph above is described only in a free-text annotation, but various approaches can be taken to formally represent Assertion semantics. SEPIO leaves such modeling choices to the implementer. In the Monarch Initiative dataset, Assertion semantics are represented using the OBAN model, which reifies an RDF triple that formally expresses the meaning expressed in an Assertion as a node in the graph. The node, which OBAN calls an 'Association', represents a Proposition can be linked to any number of Assertions that put it forth, and to any number of Evidence Lines that support or refute it. See the page [here] for further discussion of this topic, including a depiction of how Monarch uses SEPIO and OBAN models to formalize the semantics of the CIViC Assertion above.