Sequence Variant Classification Frameworks

Sequence variant classification frameworks are structured rule sets that translate scattered evidence about a DNA variant into a single, reproducible significance category. Rather than relying on individual judgement, they specify which kinds of evidence count, how strongly each weighs, and how the weighted evidence combines into tiers such as pathogenic, likely pathogenic, uncertain significance, likely benign, and benign.

Definition

A sequence variant classification framework is a codified set of evidence criteria and combining rules that assigns a DNA sequence variant to a defined significance category using population, computational, functional, and segregation evidence.

Scope

The entry covers the logic of evidence-based variant classification: the lines of evidence used, the tiered output, the dominant frameworks for germline and somatic settings, and the practical challenges of consistency and reclassification. It is a methodological topic, not clinical guidance for any individual variant.

Core questions

• What categories of evidence are admissible, and how is each weighted?
• How do weighted criteria combine into a final significance tier?
• How do germline and somatic frameworks differ in their goals and outputs?
• How is classification kept consistent across laboratories, and when should a variant be reclassified?

Key concepts

• Five-tier germline classification
• Evidence criteria and strength levels
• Population frequency evidence
• Computational (in silico) prediction
• Functional and segregation evidence
• Variant of uncertain significance
• Reclassification over time

Mechanisms

A framework defines discrete evidence criteria — for example, absence or rarity in large population databases, computational predictions of damage, experimental functional data, de novo occurrence, and co-segregation with disease in families — and assigns each a direction (toward pathogenic or benign) and a strength (such as supporting, moderate, strong, very strong). The criteria met by a given variant are then combined according to explicit rules to yield one of five germline tiers (Richards et al., 2015). Population reference resources such as large constraint databases supply much of the frequency evidence and quantify how tolerant each gene is to variation (Karczewski et al., 2020), while computational prioritisation tools rank candidate variants before formal classification (Eilbeck et al., 2017). The cancer framework adapts this logic to clinical actionability rather than heritable risk, ranking somatic variants by therapeutic, prognostic, and diagnostic significance (Li et al., 2017).

Clinical relevance

Classification frameworks determine how variant findings are reported to clinicians and patients, and understanding their tiers and limits is central to appraising a molecular report. The entry explains the reasoning behind the categories; it does not provide variant-specific interpretation or counselling, which require qualified clinical input.

Evidence & guidelines

The principal germline reference is the 2015 ACMG/AMP guideline (Richards et al., 2015), and the principal somatic reference is the 2017 AMP/ASCO/CAP guideline (Li et al., 2017). Population constraint data (Karczewski et al., 2020) and reviews of variant prioritisation (Eilbeck et al., 2017) describe the evidence sources these frameworks draw upon.

History

Before standardised criteria, laboratories classified variants with inconsistent ad hoc rules, producing conflicting calls for the same variant. The 2015 ACMG/AMP guideline consolidated germline practice into a shared five-tier scheme, and the 2017 AMP/ASCO/CAP guideline did the same for cancer; expanding population databases later sharpened the frequency-based criteria.

Debates

How should discordant classifications between laboratories be resolved?

The same variant is sometimes classified differently by different labs because criteria are applied with judgement; data sharing and explicit specification of criteria strengths are proposed to reduce discordance, but residual subjectivity remains.

Seminal works

• richards-2015
• li-2017

Frequently asked questions

What are the five categories in the ACMG/AMP germline framework?

Pathogenic, likely pathogenic, uncertain significance, likely benign, and benign — assigned by combining weighted lines of evidence according to defined rules.

Why might the same variant be classified differently for a heritable disease versus a cancer?

Germline frameworks ask whether a variant causes inherited disease, whereas the cancer framework ranks somatic variants by clinical actionability such as therapeutic relevance, so the goals and outputs differ.