civic

CIViC

⚠️ Since this loader produces statements, ontology and vocabulary data should be loaded first

Loads statements into GraphKB using the CIViC GraphQL API.

node bin/load.js civic

About

The variant.js module handles all parsing/processing of variant records in CIViC for entry into GraphKB. The corresponding tests are under test/civic.variant.test.js.

In general this loader attempts to update existing records based on their CIViC evidence ID, however when the mapping from CIViC evidence record to GraphKB statement is not 1 to 1 then the statement is sometimes recreated and the old version soft-deleted instead since we cannot resolve tracking between the 2 for these cases.

Trusted Curators

By default this loader only loads accepted (reviewed and approved) evidence items from CIViC. However, sometimes items can take a while to be reviewed. To mitigate this problem we allow a list of "trusted curators" for which we can load statements in the submitted state. These will be loaded with a GraphKB status of "pending". The IDs that should be passed are the CIViC user IDs. For example

node bin/load.js civic --trustedCurators 123 124

Mapping Objects from CIViC to GraphKB

In general CIViC (v2.2) and GraphKB have a lot in common so this mapping can be fairly straightforward. In CIViC Evidence Items are similar to GraphKB statements. However, there are a couple of key differences

• In CIViC a list of therapies (formely drugs) can be given as "substitutes", in GraphKB these are separate statements
• In CIViC, while evidence items are assigned a unique molecular profile, each profile can be associated with any number of variants using conditional expression with AND/OR/NOT operators. In GraphKB, statements are associated with any number of variants as implicit necessary conditions (AND operator). Alternative conditions (OR operator) are supported by creating separate statements like with therapy substitutes. Negative conditions (NOT operator) are yet to be supported
• GraphKB has a separate field for "appliesTo", this is the target of the statement
• Each Evidence Item in CIViC is associated with 1 PubMed article. In GraphKB a statement may have 1-n associated articles.

A mapping between CIViC and GraphKB for Evidence Item vs Statement fields is given below

GraphKB Field	CIViC Field
conditions	disease
conditions	variant
conditions	therapy
relevance	evidence type
relevance	evidence direction
relevance	significance
evidence	source

The current implementation of the relevance mapping from CIViC to GraphKB is summarized below

CIViC Evidence Type	CIViC Evidence Direction	CIViC Significance	GraphKB Relevance
Diagnostic	Supports	Negative	opposes diagnosis
Diagnostic	Supports	Positive	favours diagnosis
Functional	Supports	Dominant Negative	dominant negative
Functional	Does Not Support	Dominant Negative	no dominant negative
Functional	Supports	Gain of Function	gain of function
Functional	Does Not Support	Gain of Function	no gain of function
Functional	Supports	Loss of Function	loss of function
Functional	Supports	Neomorphic	neomorphic
Functional	Does Not Support	Neomorphic	no neomorphic
Functional	Supports	Unaltered function	Unaltered function
Oncogenic	Supports	Oncogenicity	likely oncogenic
Oncogenic	Does Not Support	Oncogenicity	likely benign
Predictive	Supports	Adverse Response	adverse response
Predictive	Supports	Reduced Sensitivity	reduced Sensitivity
Predictive	Supports	Resistance	resistance
Predictive	Does Not Support	Resistance	no resistance
Predictive	Supports	Sensitivity/Response	sensitivity
Predictive	Does Not Support	Sensitivity/Response	no response
Predisposing	Supports	Predisposition	likely predisposing
Predisposing	Supports	Protectiveness	likely protective
Prognostic	Supports	Better Outcome	favourable prognosis
Prognostic	Supports	Poor Outcome	unfavourable prognosis

Deprecated CIViC terms still supported for relevance mapping are summarized below

CIViC Evidence Type	CIViC Evidence Direction	CIViC Significance	GraphKB Relevance
Predisposing	Supports	Likely Pathogenic	likely pathogenic
Predisposing	Supports	Pathogenic	pathogenic
Predisposing	Supports	Positive	Predisposing
Predisposing	Supports	Uncertain Significance	likely predisposing