Creation of a knowledgebase of high quality assertions of the clinical actionability of somatic variants in cancer

点击翻译按钮获取中文摘要

Project Summary Interpretation of the clinical significance of somatic variants in cancer remains a major challenge in cancer diagnosis, prognosis and treatment prediction. The Clinical Genome Resource (ClinGen) has established extensive infrastructure including tools, web resources, procedures, and expert panel templates to help communities of experts establish the clinical relevance of genes and variants. However, ClinGen’s effort is almost exclusively focused on the interpretation of germline variants and their role in heritable phenotypes, leaving a significant gap in clinical interpretation of somatic variants in cancer. Concurrently the Global Alliance for Genomics in Health (GA4GH) has begun to develop standards for genomic data interoperability. The Clinical Interpretation of Variants in Cancer (CIViC) resource provides a sophisticated variant knowledge curation interface. In this application, we propose to create a knowledgebase of high quality assertions of the clinical significance of somatic variants in cancer that utilizes the CIViC platform, adapts the procedures of ClinGen to somatic variant interpretation and implements the interoperability standards of the GA4GH. This effort is needed to broadly enable research and clinical translation involving the use of somatic cancer variant knowledge as it relates to oncogenicity, diagnosis, prognosis and therapeutic response. By adapting the ClinGen germline model, we will establish processes to engage the expert community and facilitate the creation of Somatic Cancer Variant Curation Expert Panels (SC-VCEPs). Formation of these SC-VCEPs will support creation of a ClinGen Somatic Knowledgebase of clinical cancer variant assertions curated and approved by experts. SC-VCEPs will be the primary drivers of curation and domain specific guideline creation. We will adopt and guide ongoing development of several emerging standards that enable the Findable, Accessible, Interoperable, and Reusable (FAIR) principles for genomic knowledge sharing. Specifically, we will adopt the GA4GH Variation Representation Specification (VRS) and associated genomic knowledge framework of the GA4GH Genomic Knowledge Standards (GKS) Work Stream. We will also use our expert-driven curation activities to inform and develop curation and minimal information standards. Finally, we will use natural language processing (NLP) to accelerate a set of defined human knowledge curation tasks that currently limit the rate of human curation. Specifically, NLP will be used to (1) streamline curator activities through integration of text-mined data directly into CIViC; (2) prioritize papers based on their likely evidence level such as clinical trials or case reports; (3) identify and match a larger variety of cancer variant types through a variant hierarchy system; and (4) automate simple but time-consuming tasks such as enforcing consistent synonym usage. Our ultimate goal is to support a large community of domain-specific expert panels working together to create a public knowledgebase of cancer variants and their clinical relevance.