Task

regulatory-variant scoring

Regulatory-variant scoring is assessed against DART-Eval quantitative-trait-locus effect annotations.

Sources

Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization · Methods: Zero-Shot Variant Effect Prediction; Pretraining Data; cached text lines 50–52, 85–89

1 evaluation · 1 metric row

At a glance

Inputs, training, access and other details

Explanatory profile: limited source coverage · Automated source review, 2026-09-16. 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
Datasets	DNase-sensitivity QTLs and chromatin-accessibility QTLs used in DART-Eval. Sources Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization · Methods: Zero-Shot Variant Effect Prediction; Pretraining Data; cached text lines 50–52, 85–89
Splits	The paper describes chromosome-separated pretraining partitions; evaluation is zero-shot variant scoring. Sources Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization · Methods: Zero-Shot Variant Effect Prediction; Pretraining Data; cached text lines 50–52, 85–89
Metrics	Predicted allele scores are compared with observed effects; correlation analysis is restricted to variants with significant observed effects. Sources Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization · Methods: Zero-Shot Variant Effect Prediction; Pretraining Data; cached text lines 50–52, 85–89
Baselines	Figure 4 compares with DNA-language-model scores taken from DART-EVAL and includes an ARSENAL No Prior ablation. These are imported zero-shot likelihood-scoring comparisons, separate from the paper’s supervised ChromBPNet experiments. Sources Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization · Results: zero-shot regulatory-QTL scoring, Fig.4 caption; Methods: Zero-shot Variant Effect Prediction
Leakage controls	The variant experiment uses no variant-annotation supervision and evaluates significant-effect QTLs. The zero-shot methods do not document removal of these QTL loci from sequence pretraining; chromosome splits described for pretraining and supervised models are not an explicit QTL-overlap audit. · Not reported in inspected sources Sources Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization · Zero-shot Variant Effect Prediction; pretraining partitions; Supervised Model Training; Fig.4
Uncertainty	The zero-shot results and Figure 4 caption do not specify a QTL resampling unit, replicate count or confidence-interval procedure. Variability reported in the supervised ChromBPNet comparison does not establish uncertainty for these zero-shot correlations. · Not reported in inspected sources Sources Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization · Zero-shot regulatory-QTL results and Fig.4 caption; Methods: Zero-shot Variant Effect Prediction
Entity type	Paper-specific computational evaluation protocol. Sources Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization · Methods: Zero-Shot Variant Effect Prediction; Pretraining Data; cached text lines 50–52, 85–89
Organisms	Human. Sources Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization · Methods: Zero-Shot Variant Effect Prediction; Pretraining Data; cached text lines 50–52, 85–89
Assays	DNase-sensitivity and chromatin-accessibility QTL annotations. Sources Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization · Methods: Zero-Shot Variant Effect Prediction; Pretraining Data; cached text lines 50–52, 85–89
Allowed inputs	Reference/alternate DNA sequences at the evaluated variants. Sources Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization · Methods: Zero-Shot Variant Effect Prediction; Pretraining Data; cached text lines 50–52, 85–89
Adaptation	Zero-shot variant scoring; pretraining chromosome partitions are separate from the QTL evaluation. Sources Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization · Methods: Zero-Shot Variant Effect Prediction; Pretraining Data; cached text lines 50–52, 85–89

How it works

How it worksComputational evaluation flow

Conceptual summary of the cited evaluation; exact task configuration and source version remain part of the protocol.

Sources

Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization · Methods: Zero-Shot Variant Effect Prediction; Pretraining Data; cached text lines 50–52, 85–89

Evaluation methodology

DNase-sensitivity QTLs and chromatin-accessibility QTLs used in DART-Eval. The paper describes chromosome-separated pretraining partitions; evaluation is zero-shot variant scoring. Predicted allele scores are compared with observed effects; correlation analysis is restricted to variants with significant observed effects.

Sources

Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization · Methods: Zero-Shot Variant Effect Prediction; Pretraining Data; cached text lines 50–52, 85–89

Recorded evaluations

Each evaluation records what was tested and under which conditions.

ARSENAL+ChromBPNet: regulatory-variant scoring

Tested entities and results

Release 2026-09-17-d277315f7d76 · 1 evaluation · 1 metric row. Different protocols are not a single leaderboard.

Evaluation setupMetricEvidence origin

Results grouped by the exact reported evaluation
Metric and finding	Coverage and uncertainty	Evidence
ARSENAL+ChromBPNet: regulatory-variant scoring Pipeline: ARSENAL+ChromBPNetTask: regulatory-variant scoringDataset: Yoruban LCL dsQTLs Supervised ChromBPNet variant scoring with ARSENAL motif-discovery regularization Author-reported evaluation · Evaluation metadata: needs review
0.896 AUROC Unit: fraction · Direction: unknown	Uncertainty: ±0.016 Scored: Not reported · Eligible: Not reported	source checkedShort-Context Regulatory DNA Language Models with Motif-Discovery Regularization · Table 1, Yoruban LCL dsQTLs section, ARSENAL+ChromBPNet row, AUROC column Source checking is not independent reproduction.

Papers and result coverage

Last literature check: 2026-09-17. Primary-paper discovery and source inspection. Source-checked results are not independently reproduced experiments.

Paper or primary resource	Version	Reference
Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization	preprint version in PMC	Read source DOI: 10.64898/2026.02.05.703637

What is still missing

These are trained hybrid models, not zero-shot ARSENAL scores.
Keep QTL cohorts and Pearson/Spearman/AUROC separate; verify printed ± definition before plotting error bars.

Search and extraction details

primary comparison table screened

Searches

"PMC12889687"

Evidence locations

Table 1
Methods: Pretraining Data; Supervised Model Training

Strengths and limitations

Strengths and considerations

No source-reviewed explanatory claims are recorded here yet.

Limitations and conditions

Correlation is calculated only on significant-effect variants, changing the evaluated population. Zero-shot label use does not establish that their genomic contexts were absent from pretraining.
Sources
Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization · Methods: Zero-Shot Variant Effect Prediction; Pretraining Data; cached text lines 50–52, 85–89

Profile review details

Relevant full-paper computational evaluation sections, tables/captions and cited supplementary task passages were reviewed. Reporting omissions are scoped to the inspected sources. Original numerical results are unchanged.

Stable record: reported-task-b9199a30a0bcb2

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.

17 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 evaluation; exact task configuration and source version remain part of the protocol. Individual claims	Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization Original source ↗ Methods: Zero-Shot Variant Effect Prediction; Pretraining Data; cached text lines 50–52, 85–89 Version: preprint version in PMC Retrieved: 2026-09-16T10:41:06Z	source checked automated source review · 2026-09-16 Audit details Relevant full-paper computational evaluation sections, tables/captions and cited supplementary task passages were reviewed. Reporting omissions are scoped to the inspected sources. Original numerical results are unchanged. Field: `attributes.profile.diagram.caption` Source artifact SHA-256: `4a264956e47fc633aaff6573aac368dc691dd5de709b27c7421c078608ff542a` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Diagram steps ["Input: Reference/alternate DNA sequences at the evaluated variants.","Evaluation: Zero-shot variant scoring; pretraining chromosome partitions are separate from the QTL evaluation.","Readout: Predicted allele scores are compared with observed effects; correlation analysis is restricted to variants with significant observed effects."] Individual claims	Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization Original source ↗ Methods: Zero-Shot Variant Effect Prediction; Pretraining Data; cached text lines 50–52, 85–89 Version: preprint version in PMC Retrieved: 2026-09-16T10:41:06Z	source checked automated source review · 2026-09-16 Audit details Relevant full-paper computational evaluation sections, tables/captions and cited supplementary task passages were reviewed. Reporting omissions are scoped to the inspected sources. Original numerical results are unchanged. Field: `attributes.profile.diagram.steps` Source artifact SHA-256: `4a264956e47fc633aaff6573aac368dc691dd5de709b27c7421c078608ff542a` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Diagram title Computational evaluation flow Individual claims	Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization Original source ↗ Methods: Zero-Shot Variant Effect Prediction; Pretraining Data; cached text lines 50–52, 85–89 Version: preprint version in PMC Retrieved: 2026-09-16T10:41:06Z	source checked automated source review · 2026-09-16 Audit details Relevant full-paper computational evaluation sections, tables/captions and cited supplementary task passages were reviewed. Reporting omissions are scoped to the inspected sources. Original numerical results are unchanged. Field: `attributes.profile.diagram.title` Source artifact SHA-256: `4a264956e47fc633aaff6573aac368dc691dd5de709b27c7421c078608ff542a` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Datasets DNase-sensitivity QTLs and chromatin-accessibility QTLs used in DART-Eval. Individual claims	Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization Original source ↗ Methods: Zero-Shot Variant Effect Prediction; Pretraining Data; cached text lines 50–52, 85–89 Version: preprint version in PMC Retrieved: 2026-09-16T10:41:06Z	source checked automated source review · 2026-09-16 Audit details Relevant full-paper computational evaluation sections, tables/captions and cited supplementary task passages were reviewed. Reporting omissions are scoped to the inspected sources. Original numerical results are unchanged. Field: `attributes.profile.facts.0.value` Source artifact SHA-256: `4a264956e47fc633aaff6573aac368dc691dd5de709b27c7421c078608ff542a` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Splits The paper describes chromosome-separated pretraining partitions; evaluation is zero-shot variant scoring. Individual claims	Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization Original source ↗ Methods: Zero-Shot Variant Effect Prediction; Pretraining Data; cached text lines 50–52, 85–89 Version: preprint version in PMC Retrieved: 2026-09-16T10:41:06Z	source checked automated source review · 2026-09-16 Audit details Relevant full-paper computational evaluation sections, tables/captions and cited supplementary task passages were reviewed. Reporting omissions are scoped to the inspected sources. Original numerical results are unchanged. Field: `attributes.profile.facts.1.value` Source artifact SHA-256: `4a264956e47fc633aaff6573aac368dc691dd5de709b27c7421c078608ff542a` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Adaptation Zero-shot variant scoring; pretraining chromosome partitions are separate from the QTL evaluation. Individual claims	Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization Original source ↗ Methods: Zero-Shot Variant Effect Prediction; Pretraining Data; cached text lines 50–52, 85–89 Version: preprint version in PMC Retrieved: 2026-09-16T10:41:06Z	source checked automated source review · 2026-09-16 Audit details Relevant full-paper computational evaluation sections, tables/captions and cited supplementary task passages were reviewed. Reporting omissions are scoped to the inspected sources. Original numerical results are unchanged. Field: `attributes.profile.facts.10.value` Source artifact SHA-256: `4a264956e47fc633aaff6573aac368dc691dd5de709b27c7421c078608ff542a` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Metrics Predicted allele scores are compared with observed effects; correlation analysis is restricted to variants with significant observed effects. Individual claims	Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization Original source ↗ Methods: Zero-Shot Variant Effect Prediction; Pretraining Data; cached text lines 50–52, 85–89 Version: preprint version in PMC Retrieved: 2026-09-16T10:41:06Z	source checked automated source review · 2026-09-16 Audit details Relevant full-paper computational evaluation sections, tables/captions and cited supplementary task passages were reviewed. Reporting omissions are scoped to the inspected sources. Original numerical results are unchanged. Field: `attributes.profile.facts.2.value` Source artifact SHA-256: `4a264956e47fc633aaff6573aac368dc691dd5de709b27c7421c078608ff542a` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Baselines Figure 4 compares with DNA-language-model scores taken from DART-EVAL and includes an ARSENAL No Prior ablation. These are imported zero-shot likelihood-scoring comparisons, separate from the paper’s supervised ChromBPNet experiments. Individual claims	Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization Original source ↗ Results: zero-shot regulatory-QTL scoring, Fig.4 caption; Methods: Zero-shot Variant Effect Prediction Version: preprint version in PMC Retrieved: 2026-09-16T10:41:06Z	source checked automated source review · 2026-09-16 Audit details Relevant full-paper computational evaluation sections, tables/captions and cited supplementary task passages were reviewed. Reporting omissions are scoped to the inspected sources. Original numerical results are unchanged. Field: `attributes.profile.facts.3.value` Source artifact SHA-256: `4a264956e47fc633aaff6573aac368dc691dd5de709b27c7421c078608ff542a` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Leakage controls The variant experiment uses no variant-annotation supervision and evaluates significant-effect QTLs. The zero-shot methods do not document removal of these QTL loci from sequence pretraining; chromosome splits described for pretraining and supervised models are not an explicit QTL-overlap audit. Individual claims	Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization Original source ↗ Zero-shot Variant Effect Prediction; pretraining partitions; Supervised Model Training; Fig.4 Version: preprint version in PMC Retrieved: 2026-09-16T10:41:06Z	unreported automated source review · 2026-09-16 Audit details Relevant full-paper computational evaluation sections, tables/captions and cited supplementary task passages were reviewed. Reporting omissions are scoped to the inspected sources. Original numerical results are unchanged. Field: `attributes.profile.facts.4.value` Source artifact SHA-256: `4a264956e47fc633aaff6573aac368dc691dd5de709b27c7421c078608ff542a` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Uncertainty The zero-shot results and Figure 4 caption do not specify a QTL resampling unit, replicate count or confidence-interval procedure. Variability reported in the supervised ChromBPNet comparison does not establish uncertainty for these zero-shot correlations. Individual claims	Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization Original source ↗ Zero-shot regulatory-QTL results and Fig.4 caption; Methods: Zero-shot Variant Effect Prediction Version: preprint version in PMC Retrieved: 2026-09-16T10:41:06Z	unreported automated source review · 2026-09-16 Audit details Relevant full-paper computational evaluation sections, tables/captions and cited supplementary task passages were reviewed. Reporting omissions are scoped to the inspected sources. Original numerical results are unchanged. Field: `attributes.profile.facts.5.value` Source artifact SHA-256: `4a264956e47fc633aaff6573aac368dc691dd5de709b27c7421c078608ff542a` Hash scope: Hash scope not separately documented; inspect source record 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

2 source records and release history

Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization · Original source · preprint version in PMC
Short-Context Regulatory DNA Language Models with Motif-Discovery Regularization · Original source · preprint version in PMC

Download this release

Technical metadata and extraction receipts

Stable ID: reported-task-b9199a30a0bcb2

areas: dna-genomes
tasks: regulatory-variant scoring
entity level: task
version: Not reported
task: regulatory-variant scoring
scope note: Paper-specific evaluation task; protocol completeness requires further extraction.
benchmark research: review date: 2026-09-17; status: primary_comparison_table_screened; primary sources: expansion-p3-arsenal-regulatory-dna-2026; inspected locators: Table 1; Methods: Pretraining Data; Supervised Model Training; searched queries: "PMC12889687"; gaps: These are trained hybrid models, not zero-shot ARSENAL scores.; Keep QTL cohorts and Pearson/Spearman/AUROC separate; verify printed ± definition before plotting error bars.; claim scope: Primary-paper discovery and source inspection. Source-checked results are not independently reproduced experiments.
historical missing metadata: protocol version: not_reported_in_legacy_extract; split: not_reported_in_legacy_extract
metadata review scope: historical_missing_metadata preserves the original discovery state. Current descriptive evidence and missingness are recorded in profile.facts; numerical-result review is separate.
legacy kinds: benchmark
entity classification: review date: 2026-09-17; rationale: This source-scoped record identifies the biological prediction task and holds its paper context. Preserve the existing task identity; exact split, model adaptation and scoring remain in linked evaluations or separate protocol records.; source ids: arsenal-regulatory-dna-2026; source locator: Methods: Zero-Shot Variant Effect Prediction; Pretraining Data; cached text lines 50–52, 85–89; ambiguities: A paper- or suite-specific task may constrain some inputs or metrics; that alone does not make it interchangeable with a complete versioned protocol. No protocol equivalence is inferred.; Some legacy profile Entity type facts use the generic phrase computational evaluation protocol. That boilerplate is not sufficient to establish a single fixed protocol identity or to merge this task with another protocol record.

Related records

dataset: Yoruban LCL dsQTLs
benchmark: ARSENAL+ChromBPNet: regulatory-variant scoring

Suggest a correction