Abstract 1193: Enhancing pediatric cancer variant curation and representation through standardized classification and automation
Bibliographic record
Abstract
Abstract Childhood cancers present unique challenges for variant interpretation in a clinical context due to their rarity, low mutation burden, diversity of molecular alterations, and heterogeneity among patients. Consequently, genes and variants associated with childhood tumors are under-represented in public cancer databases and knowledgebases. A focused effort is needed for the structured curation of genetic variant-level data to document diagnostic, prognostic, and therapeutic biomarkers for childhood cancers. The Pediatric Cancer Curation Advancement Subcommittee (PCCAS), a collaboration between the Clinical Interpretations of Variants in Cancer knowledgebase (CIViC; civicdb.org), the ClinGen Somatic Pediatric Cancer Taskforce, Disease Ontology (DO; disease-ontology.org) and CIViCmine (bionlp.bcgsc.ca/civicmine/), is addressing this challenge through enhanced curation, tagging, and automation. PCCAS created a pediatric specific curation standard operating procedure (SOP) to harmonize pediatric evidence entered in CIViC. Our SOP provides general guidance and considerations to define and classify childhood cancers and to represent childhood cancer evidence on a spectrum of age-related incidence and presentation. For instance, pediatric evidence in CIViC is now tagged using Human Phenotype Ontology (HPO) age of onset terms, allowing pediatric evidence to be easily searched, tracked, and sorted. We also initiated the addition of new age of onset terms to enhance the granularity of these tags. WHO ICD-O nomenclature has been chosen for pediatric disease classification in CIViC. ICD-O provides updated terminology including specific genetic subtypes, which are important in pediatric cancers where their underlying molecular profiles often define the disease. To aid curator selection of disease, we verified pediatric relevant ICD-O terms inclusion in DO and restructured DO disease hierarchies to ensure proper mapping. CIViC highlights our pediatric cancer initiative in multiple areas including a homepage feature linking directly to a dedicated pediatric advanced search that returns all evidence tagged with pediatric or young adult age of onset. Most importantly, our childhood specific SOP and initiatives are included in all ClinGen Somatic Cancer and CIViC training sessions for consistent implementation. CIViCmine supports CIViC by using natural language processing to identify important cancer biomarkers in the literature. To better identify pediatric biomarkers, we are adapting and refining CIViCmine to use MeSH terms and other approaches to enhance accuracy in the identification of childhood evidence in both the literature and CIViC. In conclusion, implementation of these procedures, features, and automation are pushing to make childhood cancer variant evidence more accessible and interpretable. Citation Format: Jason Saliba, Jake Lever, Kilannin Krysiak, Arpad Danos, Alex Wagner, Heather E. Williams, Laveniya Satgunaseelan, David Meredith, Cameron J. Grisdale, Chimene Kesserwan, Jianling Ji, Shruti Rao, Catherine Cottrell, Alanna Church, Mark Evans, Yasmina Jaufeerally-Fakim, Lynn M. Schriml, Angshumoy Roy, Gordana Raca, Malachi Griffith, Obi L. Griffith. Enhancing pediatric cancer variant curation and representation through standardized classification and automation [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 1193.
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 | 0.033 | 0.080 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.003 |
| Bibliometrics | 0.013 | 0.008 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.008 | 0.005 |
| Open science | 0.003 | 0.009 |
| Research integrity | 0.001 | 0.003 |
| Insufficient payload (model declined to judge) | 0.005 | 0.006 |
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".