Draft classification, and what ClinVar says
Two independent readings, side by side. They are expected to differ: this worksheet sees only public data and cannot apply the family, functional or case-level criteria a submitting laboratory had. A disagreement pointing in opposite directions is the one worth investigating.
Conflicts
Where the evidence disagrees with itself, ranked by how much it should change your reading. This is the part other aggregators resolve silently.
Evidence
The measurements the criteria rest on. Population frequency is an observation in people; predictor scores are model inferences. They are not equal evidence.
ClinVar submissions in full
Every assertion, with its review status and the date it was last evaluated. Stars are ClinVar's own scale; zero stars means the submitter stated no assertion criteria.
ACMG/AMP worksheet
All 28 criteria, including the ones public data cannot decide. Those are shown rather than hidden, so an empty row is never mistaken for an absence of evidence. Select any criterion to see the reasoning and what would settle it.
Sources consulted
Including any that failed. A report built from four of six sources is still useful, but only if you know which two are missing.
Every tool merges that into one label
When laboratories disagree about what a variant means, the public databases show you a single summary and throw the disagreement away. So does every tool built on them.
Dissent keeps it. It assembles the evidence from six public databases, works through the 28 criteria clinical geneticists use, and puts the contradictions first — including the ones between its own conclusion and the database's.
Pick a variant above to see it work.
What this does
Interpreting one variant means opening ClinVar, gnomAD, a predictor site, Open Targets and PubMed, then reconciling them by hand. The reconciliation is where the errors happen, and the tools that automate it hide that step behind a single label.
Dissent produces a draft ACMG/AMP worksheet with every criterion sourced, and puts the disagreements first: submitters who contradict each other, records with no stated method, assertions a decade old, predictors that split, frequencies that contradict a pathogenic claim, and germline criteria applied to somatic variants.
Pick one of the cases above to see it work.