Published-case evaluation

What was measured

172 of 1,874 cases had the reported gene in the top five (9.18%; 95% interval by paper cluster 1.21% to 20.96%). Below the 90% research target.

At least one candidate received a score in 1,873 of 1,874 cases. The reported gene was in the December 2022 knowledge base in 702 of 1,874 cases.

Genes were ranked from the phenotypes of 204 published papers in the Phenopacket Store, with the gene each paper reported as the reference. Historical HPO citations overlapped 0 of those papers.

A reported causal single-nucleotide variant matched the historical ClinVar variant key and gene in 356 of 1,358 cases with a reported variant. The check uses variants reported in the papers; patient VCFs were not available.

By publication year

Split and denominator

Reference period, used only to detect duplicates

7,304 cases

Development period

1,054 cases

Test period

1,874 cases, every one in the denominator

Fitting on patient cases

no_case_training_or_tuning

Split lock

51ffff7e933738a81750d90d43357f402054f61fd3869d89cbd0238a66944d42

Prediction hash

ac5654bab164e40bb9c4c2ea15c7f6b8db5d57dac76f3f0ba8e99ccd83cd401c

Limits of this result

• The cases are retrospective reports from the literature. Records written after diagnosis, re-reports of the same family and indirect overlap of knowledge can remain.
• A case whose reported gene was not yet in the knowledge base counts as a miss.
• Patient VCFs and independently adjudicated ACMG classifications were not available, so the project’s full pipeline was not evaluated here.
• No auditor has signed this result.

Aggregate results (JSON) · Evaluation and replay record (KR) · Project 19-1