Developing a Scalable Data Model Aligned with the Forthcoming Variant Classification Standards

ASHG 2026

Oct 20, 2026
Presentation

Christine G Preston, Neethu Shah, Gloria Cheung, Hannah Dziadzio, Mark E Mandell, Liam Mulhall, Bryan Wulf, Leslie G Biesecker, Steven M Harrison, Teri E Klein, Aleksandar Milosavljevic, Sharon E Plon, Heidi L Rehm, Matt W Wright, Alicia B Byrne, Larry Babb

Abstract
Accurate classification of genetic variant pathogenicity is central to the clinical utility of genomic medicine, yet the process is constrained by how evidence is identified, evaluated, and applied. Current guidelines published by Richards et al. in 2015 lack a structured data model to describe the process, instead relying on descriptive weights of evidence criteria codes that are combined to reach a final variant pathogenicity classification. The forthcoming ACMG/AMP/CAP/ClinGen Sequence Variant Classification (SVCv4) Standards will introduce a more granular, decision-flow framework for variant classification, using new evidence codes and defined point values for specific evidence types. Implementing this framework requires infrastructure to support consistent use across users, tools, and settings.

 

Here we present a draft foundational data model representing the SVCv4 evidence-to-classification workflow, reflecting its hierarchical structure, defined decision logic, and computational format. Developed as a ClinGen driver project within the Global Alliance for Genomics and Health (GA4GH), this model applies, integrates with, and extends the GA4GH Genomic Knowledge Standards, and will be applied within the ClinGen Variant Curation Interface and Evidence Repository.

 

This data model enables evidence to be captured consistently and decision logic to be traced from individual evidence elements through rule application to final classification outcomes. It provides a framework for developing SVCv4-aligned curation interfaces, expert knowledge resources, and evidence databases and, by establishing a common schema, facilitates interoperability and data exchange across platforms. This will allow groups to customize evidence application to their gene-disease context — such as defining and consistently applying criteria for in silico predictors and population evidence. It also establishes a foundation for scalable, computational approaches to variant classification and for future integration with clinical workflows and electronic health records. This work supports more reproducible, transparent, and consistent variant classification, strengthening genomic medicine as a clinically useful tool for patient care.