Dissent

the scale of it

Every figure here is counted from ClinVar's own submission release — one row per submitted record — not estimated, and not sampled. The script that counts them is analysis/clinvar_baseline.py.

Where disagreement concentrates

Genes ranked by the share of their variants that more than one laboratory has classified differently. Only genes with enough classified variants for the figure to mean anything are included.

Where it goes wrong outright

Variants that one laboratory calls disease-causing and another calls harmless. Not a rate — a list, ordered by how well reviewed the contradiction is. Select any of them to open the full worksheet.

Look up a gene

How often laboratories disagree about variants in one gene, and where that places it among all genes measured.

How many laboratories weigh in

Most variants carry a single record, which is why disagreement is invisible until you look for it: there is usually nothing to disagree with.

What these numbers are and are not. They count classification labels that differ between records for the same variant. Some of that spread is genuine disagreement; some reflects records written for different conditions, or at different times as evidence accumulated. That is exactly why a per-variant report has to show the records rather than a single summary — a rate alone cannot tell you which kind you are looking at.