Protocol

Supervised distance-balanced eQTL causality (AlphaGenome paper)

Can the model distinguish putatively causal expression variants from low-probability variants after balancing distance to the target gene?

Sources (3)

AlphaGenome Nature 2026 supplementary comparison tables; AlphaGenome Nature 2026 supplementary methods; alphagenome: Journal full-text XML · 'Suppl Table 4 Variant performan'!A14:P14; Supplementary Methods p.30, Chromosome Splits for Variant Benchmarks; Supplementary Methods p.33, zero-shot and supervised eQTL causality; methods: pp.32–33, Expression Quantitative Trait Loci (eQTL) Variants; methods: p.30, Chromosome Splits for Variant Benchmarks; paper: Fig.1e; tables: Suppl Table 4 Variant performan; evaluation index 12; sheet rows 14

2 evaluations · 2 metric rows

At a glance

Inputs, training, access and other details

Explanatory profile: limited source coverage · Automated source review, 2026-09-17. Review applies to the cited claims; unresolved fields are listed below. Numerical results retain their own review status.

Data, procedure and scoring
Property	Description and evidence
Dataset and biological context	Reprocessed GTEx SuSiE eQTL-catalogue SNVs; positives PIP≥0.9, negatives PIP≤0.01, GENCODEv46 genes, controls downsampled within log-distance-to-TSS bins. Sources (3) AlphaGenome Nature 2026 supplementary comparison tables; AlphaGenome Nature 2026 supplementary methods; alphagenome: Journal full-text XML · 'Suppl Table 4 Variant performan'!A14:P14; Supplementary Methods p.30, Chromosome Splits for Variant Benchmarks; Supplementary Methods p.33, zero-shot and supervised eQTL causality; methods: pp.32–33, Expression Quantitative Trait Loci (eQTL) Variants; methods: p.30, Chromosome Splits for Variant Benchmarks; paper: Fig.1e; tables: Suppl Table 4 Variant performan; evaluation index 12; sheet rows 14
Split	Supervised training chromosomes are 1,4,7,8,10,13,15; validation chromosomes 2,5,11,14,17,20,22,X; test chromosomes 3,6,9,12,16,18,19,21. This is downstream evaluation separation, not a held-out reference-genome claim for the distilled backbone. The task additionally describes five-fold random-forest training excluding test chromosomes, followed by mean test auROC across those fits; exact fold membership is not enumerated. Sources (3) AlphaGenome Nature 2026 supplementary comparison tables; AlphaGenome Nature 2026 supplementary methods; alphagenome: Journal full-text XML · 'Suppl Table 4 Variant performan'!A14:P14; Supplementary Methods p.30, Chromosome Splits for Variant Benchmarks; Supplementary Methods p.33, zero-shot and supervised eQTL causality; methods: pp.32–33, Expression Quantitative Trait Loci (eQTL) Variants; methods: p.30, Chromosome Splits for Variant Benchmarks; paper: Fig.1e; tables: Suppl Table 4 Variant performan; evaluation index 12; sheet rows 14
Allowed inputs and adaptation	Absolute multimodal AlphaGenome variant features, with target-gene selection for gene-specific scores. Sources (3) AlphaGenome Nature 2026 supplementary comparison tables; AlphaGenome Nature 2026 supplementary methods; alphagenome: Journal full-text XML · 'Suppl Table 4 Variant performan'!A14:P14; Supplementary Methods p.30, Chromosome Splits for Variant Benchmarks; Supplementary Methods p.33, zero-shot and supervised eQTL causality; methods: pp.32–33, Expression Quantitative Trait Loci (eQTL) Variants; methods: p.30, Chromosome Splits for Variant Benchmarks; paper: Fig.1e; tables: Suppl Table 4 Variant performan; evaluation index 12; sheet rows 14
Metrics as reported	tissue_weighted_mean_auroc Sources (3) AlphaGenome Nature 2026 supplementary comparison tables; AlphaGenome Nature 2026 supplementary methods; alphagenome: Journal full-text XML · 'Suppl Table 4 Variant performan'!A14:P14; Supplementary Methods p.30, Chromosome Splits for Variant Benchmarks; Supplementary Methods p.33, zero-shot and supervised eQTL causality; methods: pp.32–33, Expression Quantitative Trait Loci (eQTL) Variants; methods: p.30, Chromosome Splits for Variant Benchmarks; paper: Fig.1e; tables: Suppl Table 4 Variant performan; evaluation index 12; sheet rows 14
Aggregation	auROC per tissue weighted by the tissue’s variant count; the supervised analysis additionally averages auROC across five fitted folds. Sources (3) AlphaGenome Nature 2026 supplementary comparison tables; AlphaGenome Nature 2026 supplementary methods; alphagenome: Journal full-text XML · 'Suppl Table 4 Variant performan'!A14:P14; Supplementary Methods p.30, Chromosome Splits for Variant Benchmarks; Supplementary Methods p.33, zero-shot and supervised eQTL causality; methods: pp.32–33, Expression Quantitative Trait Loci (eQTL) Variants; methods: p.30, Chromosome Splits for Variant Benchmarks; paper: Fig.1e; tables: Suppl Table 4 Variant performan; evaluation index 12; sheet rows 14
Uncertainty	Not reported for these summary-table scores. · Not reported in inspected sources Sources (3) AlphaGenome Nature 2026 supplementary comparison tables; AlphaGenome Nature 2026 supplementary methods; alphagenome: Journal full-text XML · 'Suppl Table 4 Variant performan'!A14:P14; Supplementary Methods p.30, Chromosome Splits for Variant Benchmarks; Supplementary Methods p.33, zero-shot and supervised eQTL causality; methods: pp.32–33, Expression Quantitative Trait Loci (eQTL) Variants; methods: p.30, Chromosome Splits for Variant Benchmarks; paper: Fig.1e; tables: Suppl Table 4 Variant performan; evaluation index 12; sheet rows 14
Organisms	Not extracted or verified for this record.
Assays	Not extracted or verified for this record.
Baselines	Not extracted or verified for this record.

How it works

How it worksSupervised distance-balanced eQTL causality: evaluation procedure

Conceptual summary of the cited procedure; model-specific conditions are given below.

Sources (3)

What is tested

Can the model distinguish putatively causal expression variants from low-probability variants after balancing distance to the target gene?

Sources (3)

Procedure

Train the source’s random-forest classifier on the same balanced causality dataset and score the held-out chromosome test set.

Sources (3)

Recorded evaluations

Each evaluation records what was tested and under which conditions.

Published comparisons

Explore the results reported under one evaluation protocol. Each figure keeps its source, dataset and metric together; it is not a ranking across studies.

Comparison and metric

Supervised distance-balanced eQTL causality (AlphaGenome paper)

tissue_weighted_mean_auroc (dimensionless) · Higher values are better for this metric.

Can the model distinguish putatively causal expression variants from low-probability variants after balancing distance to the target gene?

Evaluation protocol · Supervised distance-balanced eQTL causality: evaluated data subset

borzoi-ensemble: variant scores + random forest (paper Table 4) · Pipeline · Author-reported evaluation0.759224069094061
AlphaGenome absolute multimodal scores → RandomForestClassifier · Pipeline · Author-reported evaluation0.80

Source order is preserved. Plotted marks show point estimates; uncertainty, where reported, is retained in the printed values and table. Differences do not establish statistical significance.

AlphaGenome Nature 2026 supplementary comparison tables · 'Suppl Table 4 Variant performan'!L14; 'Suppl Table 4 Variant performan'!M14

Values, uncertainty and evidence

tissue_weighted_mean_auroc: original source values
Tested entity	Printed value	Uncertainty	Evidence
borzoi-ensemble: variant scores + random forest (paper Table 4) · Pipeline	0.759224069094061 dimensionless	Not reported	Author-reported evaluation · source checkedAlphaGenome Nature 2026 supplementary comparison tables · 'Suppl Table 4 Variant performan'!L14
AlphaGenome absolute multimodal scores → RandomForestClassifier · Pipeline	0.80 dimensionless	Not reported	Author-reported evaluation · source checkedAlphaGenome Nature 2026 supplementary comparison tables · 'Suppl Table 4 Variant performan'!M14

Scope and limitations

This is the AlphaGenome paper’s comparison. Source-checked scores are not independent reproductions.
Model inputs, training and inference budgets differ or remain partly unextracted. This figure does not establish a controlled architectural advantage.
Superseded and quarantined score conflicts are excluded; comparator-specific subsets remain separate.

Source transcription and grouping reviewed by automated source review on 2026-09-17. These experiments were not independently reproduced by rewire.

Tested entities and results

Release 2026-09-17-d277315f7d76 · 2 evaluations · 2 metric rows. Different protocols are not a single leaderboard.

Evaluation setupMetricEvidence origin

Results grouped by the exact reported evaluation
Metric and finding	Coverage and uncertainty	Evidence
borzoi-ensemble: variant scores + random forest (paper Table 4): Supervised distance-balanced eQTL causality Pipeline: borzoi-ensemble: variant scores + random forest (paper Table 4)Protocol: Supervised distance-balanced eQTL causality (AlphaGenome paper)Dataset subset: Supervised distance-balanced eQTL causality: evaluated data subset Train the source’s random-forest classifier on the same balanced causality dataset and score the held-out chromosome test set. Author-reported evaluation · Evaluation metadata: needs review
0.759224069094061 tissue_weighted_mean_auroc Unit: dimensionless · Direction: higher Aggregation: auROC per tissue weighted by the tissue’s variant count; the supervised analysis additionally averages auROC across five fitted folds.	Uncertainty: unreported Scored: Not reported · Eligible: Not reported	source checkedAlphaGenome Nature 2026 supplementary comparison tables · 'Suppl Table 4 Variant performan'!L14 Source checking is not independent reproduction.
AlphaGenome absolute multimodal scores → RandomForestClassifier: Supervised distance-balanced eQTL causality Pipeline: AlphaGenome absolute multimodal scores → RandomForestClassifierProtocol: Supervised distance-balanced eQTL causality (AlphaGenome paper)Dataset subset: Supervised distance-balanced eQTL causality: evaluated data subset Train the source’s random-forest classifier on the same balanced causality dataset and score the held-out chromosome test set. Author-reported evaluation · Evaluation metadata: needs review
0.80 tissue_weighted_mean_auroc Unit: dimensionless · Direction: higher Aggregation: auROC per tissue weighted by the tissue’s variant count; the supervised analysis additionally averages auROC across five fitted folds.	Uncertainty: unreported Scored: Not reported · Eligible: Not reported	source checkedAlphaGenome Nature 2026 supplementary comparison tables · 'Suppl Table 4 Variant performan'!M14 Source checking is not independent reproduction.

Papers and result coverage

Last literature check: 2026-09-17. Original Nature paper comparison; numerical source transcription was separately reviewed. No independent model execution.

Paper or primary resource	Version	Reference
alphagenome: Journal full-text XML	Retrieved page snapshot; no immutable publisher revision supplied	Read source
AlphaGenome Nature 2026 supplementary comparison tables	Nature version of record, 28 January 2026	Read source DOI: 10.1038/s41586-025-10014-0
AlphaGenome Nature 2026 supplementary methods	Supplement to Nature version of record, 28 January 2026; content hash pinned	Read source

What is still missing

Individual protocol input manifests, scoring counts and uncertainty remain unextracted where explicitly marked.
Subsequent studies use different datasets and are not pooled with this paper.

Search and extraction details

complete comparison extracted

Searches

AlphaGenome benchmark evaluation Nature 2026 supplementary Tables 3 4 regulatory variant prediction
AlphaGenome independent evaluation benchmark 2026 variant effects

Evidence locations

'Suppl Table 4 Variant performan'!A14:P14; Supplementary Methods p.30, Chromosome Splits for Variant Benchmarks; Supplementary Methods p.33, zero-shot and supervised eQTL causality; methods: pp.32–33, Expression Quantitative Trait Loci (eQTL) Variants; methods: p.30, Chromosome Splits for Variant Benchmarks; paper: Fig.1e; tables: Suppl Table 4 Variant performan; evaluation index 12; sheet rows 14

Strengths and limitations

Strengths and considerations

No source-reviewed explanatory claims are recorded here yet.

Limitations and conditions

Fine-mapping labels represent putative causality. Distance balancing changes the sampled population and does not establish calibration to genome-wide prevalence. Do not equate the supervised feature pipeline with the RNA-only scorer.
Sources (3)
AlphaGenome Nature 2026 supplementary comparison tables; AlphaGenome Nature 2026 supplementary methods; alphagenome: Journal full-text XML · 'Suppl Table 4 Variant performan'!A14:P14; Supplementary Methods p.30, Chromosome Splits for Variant Benchmarks; Supplementary Methods p.33, zero-shot and supervised eQTL causality; methods: pp.32–33, Expression Quantitative Trait Loci (eQTL) Variants; methods: p.30, Chromosome Splits for Variant Benchmarks; paper: Fig.1e; tables: Suppl Table 4 Variant performan; evaluation index 12; sheet rows 14
This is an author-reported protocol, not a rewire rerun. Different datasets, processing and adaptations cannot support an unrestricted leaderboard.
Sources (3)
AlphaGenome Nature 2026 supplementary comparison tables; AlphaGenome Nature 2026 supplementary methods; alphagenome: Journal full-text XML · 'Suppl Table 4 Variant performan'!A14:P14; Supplementary Methods p.30, Chromosome Splits for Variant Benchmarks; Supplementary Methods p.33, zero-shot and supervised eQTL causality; methods: pp.32–33, Expression Quantitative Trait Loci (eQTL) Variants; methods: p.30, Chromosome Splits for Variant Benchmarks; paper: Fig.1e; tables: Suppl Table 4 Variant performan; evaluation index 12; sheet rows 14

Profile review details

Primary-source transcription and separate automated source review. No human sign-off or independent experimental reproduction.

Stable record: alphagenome-2026-t4-protocol-12

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.

42 evidence rows matching the loaded filters

Claims, original sources and review scope · Release 2026-09-17-d277315f7d76
Property and statement	Original source and location	Review and provenance
Diagram caption Conceptual summary of the cited procedure; model-specific conditions are given below. Individual claims	alphagenome: Journal full-text XML Original source ↗ 'Suppl Table 4 Variant performan'!A14:P14; Supplementary Methods p.30, Chromosome Splits for Variant Benchmarks; Supplementary Methods p.33, zero-shot and supervised eQTL causality; methods: pp.32–33, Expression Quantitative Trait Loci (eQTL) Variants; methods: p.30, Chromosome Splits for Variant Benchmarks; paper: Fig.1e; tables: Suppl Table 4 Variant performan; evaluation index 12; sheet rows 14 Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: Retrieved page snapshot; no immutable publisher revision supplied Retrieved: 2026-09-16T19:53:03.009088+00:00	source checked automated source review · 2026-09-17 Audit details Primary-source transcription and separate automated source review. No human sign-off or independent experimental reproduction. Field: `attributes.profile.diagram.caption` Source artifact SHA-256: `d159b791fc6cb7b679c151727b7b05a6e5b6388b08a89b675014983d85b6df36` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact
Diagram caption Conceptual summary of the cited procedure; model-specific conditions are given below. Individual claims	AlphaGenome Nature 2026 supplementary methods Original source ↗ 'Suppl Table 4 Variant performan'!A14:P14; Supplementary Methods p.30, Chromosome Splits for Variant Benchmarks; Supplementary Methods p.33, zero-shot and supervised eQTL causality; methods: pp.32–33, Expression Quantitative Trait Loci (eQTL) Variants; methods: p.30, Chromosome Splits for Variant Benchmarks; paper: Fig.1e; tables: Suppl Table 4 Variant performan; evaluation index 12; sheet rows 14 Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: Supplement to Nature version of record, 28 January 2026; content hash pinned Retrieved: 2026-09-17T06:37:01.778972+00:00	source checked automated source review · 2026-09-17 Audit details Primary-source transcription and separate automated source review. No human sign-off or independent experimental reproduction. Field: `attributes.profile.diagram.caption` Source artifact SHA-256: `86b2e6a07543e3c201e2e157a3e9c6eb5235ff13eb8a5224f6fcb6fe1808d4c0` Hash scope: SHA-256 of exact publisher PDF bytes Format: original_pdf Inspected artifact
Diagram caption Conceptual summary of the cited procedure; model-specific conditions are given below. Individual claims	AlphaGenome Nature 2026 supplementary comparison tables Original source ↗ 'Suppl Table 4 Variant performan'!A14:P14; Supplementary Methods p.30, Chromosome Splits for Variant Benchmarks; Supplementary Methods p.33, zero-shot and supervised eQTL causality; methods: pp.32–33, Expression Quantitative Trait Loci (eQTL) Variants; methods: p.30, Chromosome Splits for Variant Benchmarks; paper: Fig.1e; tables: Suppl Table 4 Variant performan; evaluation index 12; sheet rows 14 Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: Nature version of record, 28 January 2026 Retrieved: 2026-09-17T06:31:06.887921+00:00	source checked automated source review · 2026-09-17 Audit details Primary-source transcription and separate automated source review. No human sign-off or independent experimental reproduction. Field: `attributes.profile.diagram.caption` Source artifact SHA-256: `833cb78b6ae6fe39415cfff296ac00c48d800326139f13eb307531a1cc133154` Hash scope: SHA-256 of exact retrieved original artifact bytes Format: xlsx Inspected artifact
Diagram steps ["Create distance-balanced fine-mapping labels","Hold out test chromosomes","Fit multimodal random forest","Aggregate tissue auROC"] Individual claims	alphagenome: Journal full-text XML Original source ↗ 'Suppl Table 4 Variant performan'!A14:P14; Supplementary Methods p.30, Chromosome Splits for Variant Benchmarks; Supplementary Methods p.33, zero-shot and supervised eQTL causality; methods: pp.32–33, Expression Quantitative Trait Loci (eQTL) Variants; methods: p.30, Chromosome Splits for Variant Benchmarks; paper: Fig.1e; tables: Suppl Table 4 Variant performan; evaluation index 12; sheet rows 14 Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: Retrieved page snapshot; no immutable publisher revision supplied Retrieved: 2026-09-16T19:53:03.009088+00:00	source checked automated source review · 2026-09-17 Audit details Primary-source transcription and separate automated source review. No human sign-off or independent experimental reproduction. Field: `attributes.profile.diagram.steps` Source artifact SHA-256: `d159b791fc6cb7b679c151727b7b05a6e5b6388b08a89b675014983d85b6df36` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact
Diagram steps ["Create distance-balanced fine-mapping labels","Hold out test chromosomes","Fit multimodal random forest","Aggregate tissue auROC"] Individual claims	AlphaGenome Nature 2026 supplementary methods Original source ↗ 'Suppl Table 4 Variant performan'!A14:P14; Supplementary Methods p.30, Chromosome Splits for Variant Benchmarks; Supplementary Methods p.33, zero-shot and supervised eQTL causality; methods: pp.32–33, Expression Quantitative Trait Loci (eQTL) Variants; methods: p.30, Chromosome Splits for Variant Benchmarks; paper: Fig.1e; tables: Suppl Table 4 Variant performan; evaluation index 12; sheet rows 14 Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: Supplement to Nature version of record, 28 January 2026; content hash pinned Retrieved: 2026-09-17T06:37:01.778972+00:00	source checked automated source review · 2026-09-17 Audit details Primary-source transcription and separate automated source review. No human sign-off or independent experimental reproduction. Field: `attributes.profile.diagram.steps` Source artifact SHA-256: `86b2e6a07543e3c201e2e157a3e9c6eb5235ff13eb8a5224f6fcb6fe1808d4c0` Hash scope: SHA-256 of exact publisher PDF bytes Format: original_pdf Inspected artifact
Diagram steps ["Create distance-balanced fine-mapping labels","Hold out test chromosomes","Fit multimodal random forest","Aggregate tissue auROC"] Individual claims	AlphaGenome Nature 2026 supplementary comparison tables Original source ↗ 'Suppl Table 4 Variant performan'!A14:P14; Supplementary Methods p.30, Chromosome Splits for Variant Benchmarks; Supplementary Methods p.33, zero-shot and supervised eQTL causality; methods: pp.32–33, Expression Quantitative Trait Loci (eQTL) Variants; methods: p.30, Chromosome Splits for Variant Benchmarks; paper: Fig.1e; tables: Suppl Table 4 Variant performan; evaluation index 12; sheet rows 14 Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: Nature version of record, 28 January 2026 Retrieved: 2026-09-17T06:31:06.887921+00:00	source checked automated source review · 2026-09-17 Audit details Primary-source transcription and separate automated source review. No human sign-off or independent experimental reproduction. Field: `attributes.profile.diagram.steps` Source artifact SHA-256: `833cb78b6ae6fe39415cfff296ac00c48d800326139f13eb307531a1cc133154` Hash scope: SHA-256 of exact retrieved original artifact bytes Format: xlsx Inspected artifact
Diagram title Supervised distance-balanced eQTL causality: evaluation procedure Individual claims	alphagenome: Journal full-text XML Original source ↗ 'Suppl Table 4 Variant performan'!A14:P14; Supplementary Methods p.30, Chromosome Splits for Variant Benchmarks; Supplementary Methods p.33, zero-shot and supervised eQTL causality; methods: pp.32–33, Expression Quantitative Trait Loci (eQTL) Variants; methods: p.30, Chromosome Splits for Variant Benchmarks; paper: Fig.1e; tables: Suppl Table 4 Variant performan; evaluation index 12; sheet rows 14 Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: Retrieved page snapshot; no immutable publisher revision supplied Retrieved: 2026-09-16T19:53:03.009088+00:00	source checked automated source review · 2026-09-17 Audit details Primary-source transcription and separate automated source review. No human sign-off or independent experimental reproduction. Field: `attributes.profile.diagram.title` Source artifact SHA-256: `d159b791fc6cb7b679c151727b7b05a6e5b6388b08a89b675014983d85b6df36` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact
Diagram title Supervised distance-balanced eQTL causality: evaluation procedure Individual claims	AlphaGenome Nature 2026 supplementary methods Original source ↗ 'Suppl Table 4 Variant performan'!A14:P14; Supplementary Methods p.30, Chromosome Splits for Variant Benchmarks; Supplementary Methods p.33, zero-shot and supervised eQTL causality; methods: pp.32–33, Expression Quantitative Trait Loci (eQTL) Variants; methods: p.30, Chromosome Splits for Variant Benchmarks; paper: Fig.1e; tables: Suppl Table 4 Variant performan; evaluation index 12; sheet rows 14 Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: Supplement to Nature version of record, 28 January 2026; content hash pinned Retrieved: 2026-09-17T06:37:01.778972+00:00	source checked automated source review · 2026-09-17 Audit details Primary-source transcription and separate automated source review. No human sign-off or independent experimental reproduction. Field: `attributes.profile.diagram.title` Source artifact SHA-256: `86b2e6a07543e3c201e2e157a3e9c6eb5235ff13eb8a5224f6fcb6fe1808d4c0` Hash scope: SHA-256 of exact publisher PDF bytes Format: original_pdf Inspected artifact
Diagram title Supervised distance-balanced eQTL causality: evaluation procedure Individual claims	AlphaGenome Nature 2026 supplementary comparison tables Original source ↗ 'Suppl Table 4 Variant performan'!A14:P14; Supplementary Methods p.30, Chromosome Splits for Variant Benchmarks; Supplementary Methods p.33, zero-shot and supervised eQTL causality; methods: pp.32–33, Expression Quantitative Trait Loci (eQTL) Variants; methods: p.30, Chromosome Splits for Variant Benchmarks; paper: Fig.1e; tables: Suppl Table 4 Variant performan; evaluation index 12; sheet rows 14 Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: Nature version of record, 28 January 2026 Retrieved: 2026-09-17T06:31:06.887921+00:00	source checked automated source review · 2026-09-17 Audit details Primary-source transcription and separate automated source review. No human sign-off or independent experimental reproduction. Field: `attributes.profile.diagram.title` Source artifact SHA-256: `833cb78b6ae6fe39415cfff296ac00c48d800326139f13eb307531a1cc133154` Hash scope: SHA-256 of exact retrieved original artifact bytes Format: xlsx Inspected artifact
Dataset and biological context Reprocessed GTEx SuSiE eQTL-catalogue SNVs; positives PIP≥0.9, negatives PIP≤0.01, GENCODEv46 genes, controls downsampled within log-distance-to-TSS bins. Individual claims	alphagenome: Journal full-text XML Original source ↗ 'Suppl Table 4 Variant performan'!A14:P14; Supplementary Methods p.30, Chromosome Splits for Variant Benchmarks; Supplementary Methods p.33, zero-shot and supervised eQTL causality; methods: pp.32–33, Expression Quantitative Trait Loci (eQTL) Variants; methods: p.30, Chromosome Splits for Variant Benchmarks; paper: Fig.1e; tables: Suppl Table 4 Variant performan; evaluation index 12; sheet rows 14 Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: Retrieved page snapshot; no immutable publisher revision supplied Retrieved: 2026-09-16T19:53:03.009088+00:00	source checked automated source review · 2026-09-17 Audit details Primary-source transcription and separate automated source review. No human sign-off or independent experimental reproduction. Field: `attributes.profile.facts.0.value` Source artifact SHA-256: `d159b791fc6cb7b679c151727b7b05a6e5b6388b08a89b675014983d85b6df36` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact

All evidence (CSV)All evidence (JSONL)How to read the evidence

Sources and history

Release 2026-09-17-d277315f7d76 · Record review: needs review

3 source records and release history

AlphaGenome Nature 2026 supplementary comparison tables · Original source · Nature version of record, 28 January 2026
AlphaGenome Nature 2026 supplementary methods · Original source · Supplement to Nature version of record, 28 January 2026; content hash pinned
alphagenome: Journal full-text XML · Original source · Retrieved page snapshot; no immutable publisher revision supplied

Download this release

Technical metadata and extraction receipts

Stable ID: alphagenome-2026-t4-protocol-12

areas: dna-genomes
entity level: protocol
version: Nature version of record, 28 January 2026
source evaluation index: 12
source table: 4
reference levels: metric: tissue_weighted_mean_auroc; printed value: 0.5; numeric value: 0.5; source locator: Suppl Table 4 Variant performan!J14; note: Source reference quantity for relative-performance calculation, not a measured baseline run.
comparison panels: id: paper-figure-0984419a2b9147c798; title: Supervised distance-balanced eQTL causality (AlphaGenome paper); protocol id: alphagenome-2026-t4-protocol-12; dataset id: alphagenome-2026-t4-dataset-12; metric: tissue_weighted_mean_auroc; unit: dimensionless; direction: higher; result ids: alphagenome-2026-result-1fc462c097bd754a; alphagenome-2026-result-5059e5c4d8ea60db; source ids: source-alphagenome-nature2026-tables; source locator: 'Suppl Table 4 Variant performan'!L14; 'Suppl Table 4 Variant performan'!M14; context: Can the model distinguish putatively causal expression variants from low-probability variants after balancing distance to the target gene?; caveats: This is the AlphaGenome paper’s comparison. Source-checked scores are not independent reproductions.; Model inputs, training and inference budgets differ or remain partly unextracted. This figure does not establish a controlled architectural advantage.; Superseded and quarantined score conflicts are excluded; comparator-specific subsets remain separate.; review: method: automated_source_review; date: 2026-09-17
benchmark research: review date: 2026-09-17; status: complete_comparison_extracted; primary sources: evidence-official-56e5abfb5f12f1cd3b20; source-alphagenome-nature2026-tables; source-alphagenome-nature2026-supplementary-methods; inspected locators: 'Suppl Table 4 Variant performan'!A14:P14; Supplementary Methods p.30, Chromosome Splits for Variant Benchmarks; Supplementary Methods p.33, zero-shot and supervised eQTL causality; methods: pp.32–33, Expression Quantitative Trait Loci (eQTL) Variants; methods: p.30, Chromosome Splits for Variant Benchmarks; paper: Fig.1e; tables: Suppl Table 4 Variant performan; evaluation index 12; sheet rows 14; searched queries: AlphaGenome benchmark evaluation Nature 2026 supplementary Tables 3 4 regulatory variant prediction; AlphaGenome independent evaluation benchmark 2026 variant effects; gaps: Individual protocol input manifests, scoring counts and uncertainty remain unextracted where explicitly marked.; Subsequent studies use different datasets and are not pooled with this paper.; claim scope: Original Nature paper comparison; numerical source transcription was separately reviewed. No independent model execution.
legacy kinds: benchmark
entity classification: review date: 2026-09-17; rationale: The source-backed record identifies a specified evaluated procedure and its dataset/split/scoring context. Classify it as a protocol while preserving version and comparison restrictions.; source ids: source-alphagenome-nature2026-tables; source-alphagenome-nature2026-supplementary-methods; evidence-official-56e5abfb5f12f1cd3b20; source locator: 'Suppl Table 4 Variant performan'!A14:P14; Supplementary Methods p.30, Chromosome Splits for Variant Benchmarks; Supplementary Methods p.33, zero-shot and supervised eQTL causality; methods: pp.32–33, Expression Quantitative Trait Loci (eQTL) Variants; methods: p.30, Chromosome Splits for Variant Benchmarks; paper: Fig.1e; tables: Suppl Table 4 Variant performan; evaluation index 12; sheet rows 14; ambiguities: None recorded

Related records

Suggest a correction