Coordinate curation for variant knowledgebases, the CIViC approach
Year: 2017
Type: other
Abstract: Despite the availability of structured variant nomenclature and ontologies, representing the wide variety of variants observed in cancer known to have clinical relevance remains a challenge. These variants range from the most simple single nucleotide variants to complex chromosomal rearrangements at the DNA level and include additional layers of complexity when accounting for transcript-level or epigenetic alterations. As numerous groups undertake the daunting task of interpreting variants of clinical relevance, a need for representing the genomic coordinates of these variants in ways that provide resource interoperability and integration with downstream applications has become notable. Development of one of these resources, the crowd-sourced open-access knowledgebase CIViC (Clinical Interpretation of Variants in Cancer; civicdb.org), has led us to develop standard practices for representing variants at the genomic level. These standards, however, continue to be debated and developed in the CIViC interface as new variants and downstream applications are developed. Variant nomenclature such as HGVS representations provide a strong foundation for inter-operable genomic representations. HGVS nomenclature can be captured in CIViC but does not universally handle the variety of variants already represented in the CIViC knowledgebase. CIViC seeks to have a representative set of genomic coordinates for all variants in the knowledgebase and, to guide curators, maintains help documentation which can be accessed in the help section of the website at any time (https://civic.genome.wustl.edu/#/help/variants/variants-coordinates). All evidence supporting clinical interpretations of variants in CIViC are derived from the published medical literature and aggregated under the most detailed variant possible (i.e., “ALK fusion” when fusion partners are not defined by methods used by the literature source versus the more specific “EML4-ALK” whenever possible). Of particular interest is the best approach for representing challenging use-cases such as variants presented in the literature without explicit genomic and transcriptomic designations, literature results which aggregate multiple variants in the same gene, or complex variants such as fusions where different genomic breakpoints can have the same transcriptomic result. We will present our current approach to identifying genomic coordinate representations for variants in CIViC and the logic that drove these decisions.
Author: Kilannin Krysiak
Language: English
Topic: Genomics and Rare Diseases
Subfield: Genetics
Domain: Life Sciences
Open Access status: gold
OpenAlex ID: W2619788376