Introduction #
Variant classification is one of the central functions of SeqSMART.
It combines rule-based interpretation (ACMG/AMP guidelines), multi-source annotation, and expert-driven validation to deliver accurate and transparent variant pathogenicity assessments.
SeqSMART’s classification process is designed specifically for professional genomic laboratories — integrating automation with expert control. Every classification is traceable, evidence-based, and customizable according to institutional standards.
1. The Foundation: ACMG/AMP Guidelines #
SeqSMART’s classification engine is built upon the internationally recognized ACMG/AMP standards (American College of Medical Genetics and Genomics / Association for Molecular Pathology).
These guidelines provide a systematic framework for evaluating the clinical significance of sequence variants using a defined set of criteria categorized by evidence strength:
- Pathogenic Evidence: PVS, PS, PM, PP
- Benign Evidence: BS, BP
Each criterion is assigned a strength level — Very Strong, Strong, Moderate, or Supporting — based on the weight of evidence.
SeqSMART operationalizes these rules into a computational logic model that can automatically evaluate many criteria using integrated databases, population metrics, and predictive tools.
2. SeqSMART’s Interpretation Model #
SeqSMART uses a hybrid model that merges automated data-driven evaluation with expert oversight.
2.1 Automated Evaluation #
During automated analysis, the platform:
- Annotates each variant across genomic, transcript, and protein contexts.
- Extracts evidence from integrated databases (ClinVar, gnomAD, OMIM, HPO, UniProt, etc.).
- Evaluates ACMG criteria automatically when sufficient data are available.
- Generates a preliminary classification supported by evidence summaries.
This automation enables rapid and standardized interpretation across large variant sets.
2.2 Expert Validation #
Once automation is complete, expert analysts review the results through the ACMG interface.
They can:
- Inspect the applied criteria and underlying evidence.
- Adjust the strength or applicability of any rule.
- Add manual evidence or literature citations.
- Provide a final classification and rationale.
This structure preserves both the speed of automation and the accuracy of expert judgment.
3. Classification Workflow in SeqSMART #
SeqSMART’s classification pipeline consists of sequential steps that ensure data completeness and interpretive rigor.
- Annotation: All variants are enriched with functional, population, and clinical data.
- Rule Evaluation: The system assesses each ACMG rule individually based on available evidence.
- Evidence Integration: Positive and negative evidence are combined following ACMG’s hierarchical logic.
- Preliminary Classification: The platform assigns an initial class (Pathogenic, Likely Pathogenic, VUS, Likely Benign, Benign).
- Expert Review: Analysts confirm, modify, or justify the system’s results.
- Final Classification: The validated result becomes part of the case report and audit trail.
Each rule evaluation is stored transparently, allowing any classification to be reproduced or re-analyzed as databases evolve.
4. Evidence Sources #
SeqSMART integrates multiple internal and external data sources to support ACMG evaluation:
| Evidence Type | Primary Sources | Purpose |
| Population frequency | gnomAD, ExAC, 1000 Genomes | Identify benign variants through frequency thresholds |
| Functional predictions | CADD, PolyPhen, SIFT, REVEL | Evaluate potential protein impact |
| Clinical databases | ClinVar, OMIM, HGMD-like sources | Compare against previously reported pathogenic variants |
| Phenotype correlation | HPO, OMIM synopses | Match observed symptoms to candidate genes |
| Gene-level annotations | UniProt, Ensembl, RefSeq | Assess conservation, domain context, and LOF relevance |
| Literature evidence | Manual curation via PubMed links | Support expert review for rare or novel variants |
These datasets are continuously updated and version-tracked to ensure reproducibility across analyses.
5. Evidence Integration and Rule Hierarchy #
SeqSMART applies a weighted logic system aligned with ACMG’s combination rules:
- Pathogenic classification: Triggered by combinations such as 1 Very Strong + ≥1 Moderate, or 2 Strong + ≥1 Supporting.
- Benign classification: Triggered by sufficient benign evidence (e.g., BS1 + BP4).
- Variants of Uncertain Significance (VUS): Assigned when evidence is insufficient or conflicting.
The platform calculates final outcomes dynamically and displays all active rules and their weights for full transparency.
6. Transparency and Reproducibility #
SeqSMART prioritizes transparency in every classification decision.
For each variant, users can view:
- Which ACMG rules were met or unmet.
- The specific data or database entries supporting each rule.
- The source and version of evidence used.
- Analyst modifications and their justification.
All steps are recorded in the case audit log, ensuring that any classification can be traced, reviewed, or reproduced at a later date.
7. Flexibility and Customization #
Institutions may customize the classification framework to fit their internal policies or interpretation workflows.
Administrators can:
- Modify thresholds for frequency-based rules.
- Adjust rule weights or disable specific criteria.
- Add institution-specific annotations or curated data sources.
- Save and deploy custom ACMG templates for all analyses.
This flexibility allows SeqSMART to align with both clinical diagnostic workflows and research-specific variant assessment frameworks.
SeqSMART’s variant classification engine integrates comprehensive annotation, evidence-based ACMG evaluation, and expert-driven validation into a unified system.
It delivers standardized yet adaptable classifications that combine automation efficiency with the depth of expert interpretation.
SeqSMART Classification Principle:
Automate with intelligence, interpret with expertise.