Abstract 206: CIViC knowledgebase adapts to field experts and community input
Bibliographic record
Abstract
Abstract CIViC (civicdb.org) is an open access, expertly moderated knowledgebase for crowdsourcing Clinical Interpretations of Variants in Cancer. Stakeholders globally-including those in government, academia, industry and medicine-use CIViC to find and curate actionable interpretations of genomic variants in their therapeutic, prognostic, predisposing, diagnostic and functional contexts. Through engagement with curators and leaders in the field, CIViC has implemented several features including Assertions, Organizations and expanded help documentation. The foundational unit of CIViC is the Evidence Item, which describes the clinical relevance of a specific variant curated from a single published source within peer-reviewed literature or ASCO abstract. Assertions aggregate Evidence Items for a given variant-disease or variant-disease-therapy combination. In response to the 2017 AMP-ASCO-CAP guidelines and collaborations with ClinGen, Assertions were modified to integrate ACMG variant pathogenicity classifications, AMP-ASCO-CAP tier designations and associations with NCCN guidelines and FDA approvals to provide a ‘state of the field' interpretation. At present, 12 Assertions spanning eight variants have been submitted by CIViC to ClinVar with one-star submitter status (Submitter ID: 506594), and CIViC has been cited as supporting information in two variants. Assertions exemplify CIViC's responsiveness to new field guidelines, expert collaborators' recommendations and its contributions to other resources. To enhance community involvement, CIViC created Organization-attributed actions. Each action performed by a curator is tagged with their Organization. Curators may switch between Organizations if they belong to more than one. Currently, nine Organizations are recognized in CIViC, the largest being ClinGen with 79 members, 4 sub-organizations and over 24,000 actions. Organizations enable groups to prominently display and track their submissions, activity, and users. CIViC's wide adoption has necessitated the development of robust educational material. CIViC has created nine YouTube videos, one of which is linked by the NIH ITCR homepage. CIViC has migrated help documents to a stand alone site (civic.readthedocs.io) and has made over 60 page modifications since 2019. Help documentation expansion was fueled by user feedback via the CIViC interface, collaborator meetings and in-person events (Curation Jamborees). Improved documentation allows CIViC to grow at scale, unhindered by the need for direct training. CIViC's rapid adaptation to the needs of the community is derived from its open access nature, commitment to data provenance, active connection with users, and abundance of educational material. CIViC rapidly integrates the guidelines, regulatory standards and community recommendations in a freely accessible resource that is flexible enough to evolve with the dynamic field of cancer genomics. Citation Format: Lana M. Sheta, Arpad M. Danos, Jason Saliba, Kilannin Krysiak, Alex H. Wagner, Erica K. Barnell, Susanna Kiwala, Joshua F. McMichael, Adam Coffman, Shahil Pema, Lynzey Kujan, Kelsy C. Cotto, Cody Ramirez, Zachary L. Skidmore, Cameron J. Grisdale, Shruti Rao, Subha Madhaven, Malachi Griffith, Obi L. Griffith. CIViC knowledgebase adapts to field experts and community input [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 206.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.020 | 0.087 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.010 | 0.006 |
| Science and technology studies | 0.003 | 0.001 |
| Scholarly communication | 0.008 | 0.009 |
| Open science | 0.005 | 0.013 |
| Research integrity | 0.004 | 0.003 |
| Insufficient payload (model declined to judge) | 0.102 | 0.057 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".