AUPRC (ClinVar pathogenic vs benign SNVs, 3' UTR) of AVI six-model ensemble (AlphaGenome Atlas v1) on ClinVar 2025-06-15 SNVs, 3' UTR (AlphaGenome Atlas test set)
0.5 AUPRC (ClinVar pathogenic vs benign SNVs, 3' UTR)
Methods
- Tested configuration
- AVI six-model ensemble (AlphaGenome Atlas v1)
- Protocol
- ClinVar pathogenic versus benign SNVs by VEP consequence, AUPRC (AlphaGenome Atlas)
- Dataset subset
- ClinVar 2025-06-15 SNVs, 3' UTR (AlphaGenome Atlas test set)
- Procedure
- ClinVar pathogenic versus benign SNVs by VEP consequence, AUPRC (AlphaGenome Atlas)
- Evaluation
- avi on ClinVar 3' UTR SNVs (AlphaGenome Atlas Fig. 2A)
- Coverage
- Not reported scored / Not reported eligible
- Uncertainty
- Not reported by the source: Fig. 2A draws 95% bootstrap intervals (100 bootstraps) without printing them.
- Evidence
- Author-reported evaluation · source checkedAlphaGenome Atlas: in silico mutagenesis of the entire human genome (Cheng, Taylor, Nicolaisen et al., medRxiv 2026) · Results 'State-of-the-art interpretable genome-wide variant prioritization with AVI' paragraph 1 (page 5), citing Fig. 2A: '3’ UTR categories (0.50 vs. GPN-Star-M’s 0.18)'
A source-checked result verifies the numerical transcription, not every model or protocol detail. Evaluation metadata: source checked. Source checked does not mean independently reproduced.
Reproduction
- Split
- AVI positional test filter
- Adaptation
- None for comparators; AVI trained on gnomAD frequency labels, not ClinVar
- Scoring implementation
- AUPRC
No execution recipe has been verified for this exact configuration and evaluation. A benchmark's general instructions may use different inputs, splits or model settings.
Reproducing this published result requires matching its model configuration, data, split and scorer. Source checking or a successful smoke test does not establish score reproduction.
Evidence
Source checking verifies the cited claim or transcription. It does not establish independent reproduction.
Evidence table
Inspect claims, sources and review details
Trace each statement to its source and review. A context-only reference supports the record generally; it does not verify an individual field. Source checking does not reproduce an experiment.
One row per statement and cited source. Multiple citations are not independent evaluations. Shared locators are labelled explicitly.
1 evidence row matching the loaded filters
| Property and statement | Original source and location | Review and provenance |
|---|---|---|
| Reported result 0.50 Individual claims | AlphaGenome Atlas: in silico mutagenesis of the entire human genome (Cheng, Taylor, Nicolaisen et al., medRxiv 2026) Results 'State-of-the-art interpretable genome-wide variant prioritization with AVI' paragraph 1 (page 5), citing Fig. 2A: '3’ UTR categories (0.50 vs. GPN-Star-M’s 0.18)' Version: medRxiv preprint version 1, posted 2026-09-20; full-text PDF with supplementary figures and tables (83 pages) | source checked ["source-hash-verification","independent-cell-check","automated-source-review"] · 2026-11-01 author reported Audit detailsChecked that the value is printed inside the quoted sentence or legend entry and that the quote occurs in the pdftotext text of a 2026-11-01 retrieval; rare-disease recalls read against the paragraph on page 7. The 2026-11-01 retrieval is stamped differently from the pinned bytes (different SHA-256); re-running the extractor on it, with only the hash assertion relaxed, reproduced batch.jsonl byte for byte. Verdict in docs/reviews/2026-11-01-alphagenome-atlas-lu-brandes.md. Field: Source artifact SHA-256: Hash scope: SHA-256 of one retrieval of the medRxiv PDF. A second retrieval at 2026-10-19T12:30:54Z had the same length (15,415,633 bytes) and the same SHA-256. Extraction artifact SHA-256: |
Sources and history
Release 2026-11-02-a7f03c9e55a5 · Record review: source checked
1 source record and release history
- AlphaGenome Atlas: in silico mutagenesis of the entire human genome (Cheng, Taylor, Nicolaisen et al., medRxiv 2026) · Original source · medRxiv preprint version 1, posted 2026-09-20; full-text PDF with supplementary figures and tables (83 pages)
Technical metadata and extraction receipts
Stable ID: ag-atlas-20261019-result-clinvar-snv-utr3-avi-auprc
- metric
- auprc
- metric direction
- higher
- unit
- unitless
- printed value
- 0.50
- numeric value
- 0.50
- metric qualifier
- ClinVar pathogenic vs benign SNVs, 3' UTR
- source locator
- Results 'State-of-the-art interpretable genome-wide variant prioritization with AVI' paragraph 1 (page 5), citing Fig. 2A: '3’ UTR categories (0.50 vs. GPN-Star-M’s 0.18)'
- missing metadata
- uncertainty: reason: unreported; note: Fig. 2A draws 95% bootstrap intervals (100 bootstraps) without printing them.
- review
- method: source-hash-verification; independent-cell-check; automated-source-review; method note: Value transcribed from the prose sentence quoted in the locator; sentence asserted as an exact substring of the pdftotext -layout text (extract/build_batch.py).; reviewer: claude; reviewer note: Independent Claude review agent for PR #214, 2026-11-01; did not extract this batch; no human review claimed.; date: 2026-11-01; artifact sha256: b32d1539005ca4ea31c3dcef49b834424b1d8b5da4ce6b9fc3e1f942d39fc453; retrieval url: https://www.medrxiv.org/content/10.64898/2026.09.16.26363192v1.full.pdf; note: Checked that the value is printed inside the quoted sentence or legend entry and that the quote occurs in the pdftotext text of a 2026-11-01 retrieval; rare-disease recalls read against the paragraph on page 7. The 2026-11-01 retrieval is stamped differently from the pinned bytes (different SHA-256); re-running the extractor on it, with only the hash assertion relaxed, reproduced batch.jsonl byte for byte. Verdict in docs/reviews/2026-11-01-alphagenome-atlas-lu-brandes.md.