AI Variant Assessment Tool
AI-assisted, evidence-based ACMG/AMP variant classification, from literature to report.
A complete workflow for literature-based variant curation
Interpreting the clinical significance of a genetic variant is one of the most time-consuming and inconsistent steps in clinical genomics. It means searching the literature, reading dozens of papers, extracting structured evidence, applying ACMG/AMP criteria, and documenting the reasoning, a process that can take hours per variant and demands deep expertise at every step.
The AI Variant Assessment Tool brings that entire process into one place. For a given gene and variant, it searches leading genomic databases, identifies the publications most relevant to that specific variant, and reads each one in full, including its figures, tables, and supplementary materials. It scores each ACMG/AMP criterion, produces its own classification with a written rationale, and cross-references the result against trusted population and clinical databases. The outcome is variant curation that is faster, more transparent, and more consistent across assessments.
Everything in one platform
From discovery to a signed-off report, each part of the curation workflow is built in.
Literature Search
For a given gene and variant, the tool searches across leading genomic databases and ranks results by relevance to that exact variant, so the most informative papers rise to the top.
Deep Evidence Extraction
The AI reads each paper end to end, including figures, tables, and supplementary materials, surfacing allele counts, functional studies, segregation data, and the verbatim quotes behind every data point.
AI Assessment
A deterministic ACMG/AMP rule engine is paired with an AI reviewer that evaluates the criteria rules cannot judge mechanically, then delivers its own classification with a clear reasoning paragraph.
Trusted Cross-Referencing
Population frequency and clinical significance are pulled from trusted databases automatically and reconciled against the literature-based evidence before a classification is finalized.
Assessment Library
Every assessment is retained in a searchable library with its full evidence trail and editable results, so past work stays documented, reproducible, and easy to revisit.
Standardized Reporting
Generates ClinGen VCEP-aligned curation reports as well as lab-specific reports through customizable templates, alongside clear evidence summaries, downloadable in multiple formats.
A human at every step
Automation handles the heavy lifting, but the reviewer stays in control. Every stage has a checkpoint, so the final classification always reflects expert judgment.
Review at every checkpoint
Reviewers choose which papers to include and inspect evidence as it streams in during analysis.
Edit before classifying
The aggregated evidence summary can be edited before classification runs, and re-run anytime without re-reading papers.
Pause and resume
Assessments can be paused and picked back up later. No work is lost between sessions.
Access
This tool is currently invite-only. If your team is interested in participating, contact us at contact@datacommons.ca.