Tier C · benchmark_suite

Numeric results available

Benchmarking DNA foundation models for genomic and genetic tasks

One-time benchmark study rather than a maintained benchmark suite.

B08community_protocolnot_applicable

Result coverage

8 result rows across 6 protocols

Primary sources
primary_sources_screened
Related works
index_screen_complete
Authority readiness
discovery_only
Literature cutoff
2026-08-06
TrackNumeric coverageEvidence closureBlocker
B08-TR01
Benchmarking DNA foundation models for genomic and genetic tasks — Genomic element classification
numeric_results_available
0 results
closed_no_eligible_public_numeric_resultno_eligible_public_numeric_result
B08-TR02
Benchmarking DNA foundation models for genomic and genetic tasks — Promoter recognition
numeric_results_available
0 results
closed_no_eligible_public_numeric_resultno_eligible_public_numeric_result
B08-TR03
Benchmarking DNA foundation models for genomic and genetic tasks — TF binding-site prediction
numeric_results_available
0 results
closed_no_eligible_public_numeric_resultno_eligible_public_numeric_result
B08-TR04
Benchmarking DNA foundation models for genomic and genetic tasks — Splice-site prediction
numeric_results_available
2 results
closed_with_numeric_evidenceNR
B08-TR05
Benchmarking DNA foundation models for genomic and genetic tasks — Regulatory variant-effect prediction
numeric_results_available
4 results
closed_with_numeric_evidenceNR
B08-TR06
Benchmarking DNA foundation models for genomic and genetic tasks — Molecular functional or pathogenic variant interpretation
numeric_results_available
2 results
closed_with_numeric_evidenceNR

Protocol-level results

Choose the evaluation unit before comparing scores.

Each card keeps the four claim types inside one protocol + metric fingerprint.

Splice-site prediction · B08-TR04 · same_source_table

Benchmarking DNA foundation models for genomic and genetic tasks — Splice-site prediction

Full results →

splice donor identification · paper-defined held-out test split · paper-defined held-out evaluation

AUROCclassification · 1 results

Official baseline

No eligible claimNo qualifying result exists in this snapshot.

Original-table best

No eligible claimNo qualifying result exists in this snapshot.

Official board / challenge leader

No eligible claimNo qualifying result exists in this snapshot.

Strict cross-work SOTA

No eligible claimAn original-table best is never substituted for cross-work SOTA.

Fingerprint: PMF-v1-683B5992FB2043C2179E8E05

Splice-site prediction · B08-TR04 · same_source_table

Benchmarking DNA foundation models for genomic and genetic tasks — Splice-site prediction

Full results →

splice acceptor identification · paper-defined held-out test split · paper-defined held-out evaluation

AUROCclassification · 1 results

Official baseline

No eligible claimNo qualifying result exists in this snapshot.

Original-table best

No eligible claimNo qualifying result exists in this snapshot.

Official board / challenge leader

No eligible claimNo qualifying result exists in this snapshot.

Strict cross-work SOTA

No eligible claimAn original-table best is never substituted for cross-work SOTA.

Fingerprint: PMF-v1-26C8D1580FB51539367BC138

Molecular functional or pathogenic variant interpretation · B08-TR06 · same_source_table

Benchmarking DNA foundation models for genomic and genetic tasks — Molecular functional or pathogenic variant interpretation

Full results →

pathogenic versus common SNP · nested chromosome-based held-out test groups · paper-defined held-out evaluation

AUROCclassification · 2 results

Official baseline

No eligible claimNo qualifying result exists in this snapshot.

Original-table best

Nucleotide Transformer v20.73AUROC · fractionEvidence ↗

Official board / challenge leader

No eligible claimNo qualifying result exists in this snapshot.

Strict cross-work SOTA

No eligible claimAn original-table best is never substituted for cross-work SOTA.

Fingerprint: PMF-v1-7AA4F4DD18F6A98FD71357EB

Regulatory variant-effect prediction · B08-TR05 · same_source_table

Benchmarking DNA foundation models for genomic and genetic tasks — Regulatory variant-effect prediction

Full results →

putative causal sQTL · nested chromosome-based held-out test groups · paper-defined held-out evaluation

AUROCclassification · 1 results

Official baseline

No eligible claimNo qualifying result exists in this snapshot.

Original-table best

No eligible claimNo qualifying result exists in this snapshot.

Official board / challenge leader

No eligible claimNo qualifying result exists in this snapshot.

Strict cross-work SOTA

No eligible claimAn original-table best is never substituted for cross-work SOTA.

Fingerprint: PMF-v1-C66A17D570014B8F17BC4007

Regulatory variant-effect prediction · B08-TR05 · same_source_table

Benchmarking DNA foundation models for genomic and genetic tasks — Regulatory variant-effect prediction

Full results →

putative causal ipaQTL · nested chromosome-based held-out test groups · paper-defined held-out evaluation

AUROCclassification · 1 results

Official baseline

No eligible claimNo qualifying result exists in this snapshot.

Original-table best

No eligible claimNo qualifying result exists in this snapshot.

Official board / challenge leader

No eligible claimNo qualifying result exists in this snapshot.

Strict cross-work SOTA

No eligible claimAn original-table best is never substituted for cross-work SOTA.

Fingerprint: PMF-v1-501F080104806EB4073AEC3E

Regulatory variant-effect prediction · B08-TR05 · same_source_table

Benchmarking DNA foundation models for genomic and genetic tasks — Regulatory variant-effect prediction

Full results →

putative causal eQTL · nested chromosome-based held-out test groups · paper-defined held-out evaluation

AUROCclassification · 2 results

Official baseline

No eligible claimNo qualifying result exists in this snapshot.

Original-table best

Enformer0.77AUROC · fractionEvidence ↗

Official board / challenge leader

No eligible claimNo qualifying result exists in this snapshot.

Strict cross-work SOTA

No eligible claimAn original-table best is never substituted for cross-work SOTA.

Fingerprint: PMF-v1-9B3D9A02DE958A097E88E22C