Model comparison: functional evidence and benchmarking presents the separate 14 September 2026 frozen summary release, including the full QTL refits. The source records and stored models below retain their own source freeze dates, including the 10 September delivery and the audited 12 September eQTLGen package. They are not exact sample drilldowns for those new figures.

Start a single-study exploration in Guided Workflow, or inspect the original molecular studies below. Open Targets study IDs and the stored biobank study IDs use separate namespaces; a shared phenotype label does not establish a study match.

Gene symbol or fixed reference gene ID

Set Gene

Enter a symbol or fixed ID and click Set Gene. A small frozen index then checks the selected source before any large model artifact is opened. If the gene is present, use Find models to count only that source/gene and Show models to load the first page. GTEx eQTL contains 38,116 indexed gene IDs; the index includes models that were not testable and does not imply significance.

Source availability and provenance

Original associations, measurements, clinical records and third-party predictions are separate from the research models computed by Evo2VED. A source with no matching record does not provide a negative result.

Clinical and experimental benchmark summaries describe their original whole-study samples. ClinVar’s strict Benign versus non-Benign benchmark includes likely benign, uncertain and conflicting labels in the non-Benign group. EVEE’s original threshold comparison has only eight positives and no cross-validation. Neither supplies a clinical decision threshold for this website.