Benchmarks · genomics/dnabert-2-117m

Genomic_Benchmarks_demo_coding_vs_intergenomic_seqs genomics/genomic-benchmarks-demo-coding-vs-interg

Genomic Benchmarks accuracy genomics/genomic-benchmarks-accuracy

Drawn from: Genomic_Benchmarks_demo_coding_vs_intergenomic_seqs, Genomic_Benchmarks_human_enhancers_cohn, Genomic_Benchmarks_human_nontata_promoters

atlas-full-v1

Metric

Value

95% CI

n

Rank

Best in catalogue

Source

Overlap

Accuracy (logistic probe, mean of 3 sets) ↑ higher is better · probe

By dataset: 3 levels

dataset	Value [95% CI]	n
Coding vs intergenomic sequences (demo)	0.872 [0.868, 0.876]	25000
Human enhancers (Cohn)	0.713 [0.702, 0.723]	6948
Human non-TATA promoters	0.796 [0.787, 0.804]	9034

0.794

[0.789, 0.798]

40982

3 / 5

0.821

reproduced

overlap

Train/test overlap declared by the evaluator: DNABERT-2 multi-species genomes (135 species)

Board

Macro F1 (logistic probe, mean of 3 sets) ↑ higher is better · probe · secondary

By dataset: 3 levels

dataset	Value [95% CI]	n
Coding vs intergenomic sequences (demo)	0.872 [0.868, 0.876]	25000
Human enhancers (Cohn)	0.713 [0.702, 0.723]	6948
Human non-TATA promoters	0.795 [0.786, 0.803]	9034

0.793

[0.789, 0.798]

40982

—

0.821

reproduced

overlap

Train/test overlap declared by the evaluator: DNABERT-2 multi-species genomes (135 species)

Board

Genomic_Benchmarks_human_enhancers_cohn genomics/genomic-benchmarks-human-enhancers-cohn

Genomic Benchmarks accuracy genomics/genomic-benchmarks-accuracy

Drawn from: Genomic_Benchmarks_demo_coding_vs_intergenomic_seqs, Genomic_Benchmarks_human_enhancers_cohn, Genomic_Benchmarks_human_nontata_promoters

atlas-full-v1

Metric

Value

95% CI

n

Rank

Best in catalogue

Source

Overlap

Accuracy (logistic probe, mean of 3 sets) ↑ higher is better · probe

By dataset: 3 levels

dataset	Value [95% CI]	n
Coding vs intergenomic sequences (demo)	0.872 [0.868, 0.876]	25000
Human enhancers (Cohn)	0.713 [0.702, 0.723]	6948
Human non-TATA promoters	0.796 [0.787, 0.804]	9034

0.794

[0.789, 0.798]

40982

3 / 5

0.821

reproduced

overlap

Train/test overlap declared by the evaluator: DNABERT-2 multi-species genomes (135 species)

Board

Macro F1 (logistic probe, mean of 3 sets) ↑ higher is better · probe · secondary

By dataset: 3 levels

dataset	Value [95% CI]	n
Coding vs intergenomic sequences (demo)	0.872 [0.868, 0.876]	25000
Human enhancers (Cohn)	0.713 [0.702, 0.723]	6948
Human non-TATA promoters	0.795 [0.786, 0.803]	9034

0.793

[0.789, 0.798]

40982

—

0.821

reproduced

overlap

Train/test overlap declared by the evaluator: DNABERT-2 multi-species genomes (135 species)

Board

Genomic_Benchmarks_human_nontata_promoters genomics/genomic-benchmarks-human-nontata-promote

Genomic Benchmarks accuracy genomics/genomic-benchmarks-accuracy

Drawn from: Genomic_Benchmarks_demo_coding_vs_intergenomic_seqs, Genomic_Benchmarks_human_enhancers_cohn, Genomic_Benchmarks_human_nontata_promoters

atlas-full-v1

Metric

Value

95% CI

n

Rank

Best in catalogue

Source

Overlap

Accuracy (logistic probe, mean of 3 sets) ↑ higher is better · probe

By dataset: 3 levels

dataset	Value [95% CI]	n
Coding vs intergenomic sequences (demo)	0.872 [0.868, 0.876]	25000
Human enhancers (Cohn)	0.713 [0.702, 0.723]	6948
Human non-TATA promoters	0.796 [0.787, 0.804]	9034

0.794

[0.789, 0.798]

40982

3 / 5

0.821

reproduced

overlap

Train/test overlap declared by the evaluator: DNABERT-2 multi-species genomes (135 species)

Board

Macro F1 (logistic probe, mean of 3 sets) ↑ higher is better · probe · secondary

By dataset: 3 levels

dataset	Value [95% CI]	n
Coding vs intergenomic sequences (demo)	0.872 [0.868, 0.876]	25000
Human enhancers (Cohn)	0.713 [0.702, 0.723]	6948
Human non-TATA promoters	0.795 [0.786, 0.803]	9034

0.793

[0.789, 0.798]

40982

—

0.821

reproduced

overlap

Train/test overlap declared by the evaluator: DNABERT-2 multi-species genomes (135 species)

Board

Other Benchmarks drawn from no hub dataset you can see

Protein abundance from coding sequence (E. coli) rna-ml/coding-sequence-protein-abundance

atlas-n100-v1

Metric	Value	95% CI	n	Rank	Best in catalogue	Source	Overlap	Board
Spearman ρ (zero-shot pseudo-log-likelihood) ↑ higher is better · zero-shot	0.184	[−0.043, 0.384]	100	8 / 11	0.552	reproduced	unknown	Board
Spearman ρ (ridge probe) ↑ higher is better · probe	0.496	[0.310, 0.641]	100	8 / 11	0.688	reproduced	unknown	Board
RMSE (ridge probe) ↓ lower is better · log10 ppm · probe	1.117	[0.987, 1.243]	100	8 / 11	0.973	reproduced	unknown	Board
Pearson r (ridge probe) ↑ higher is better · probe · secondary	0.510	[0.351, 0.634]	100	—	0.665	reproduced	unknown	Board
R² (ridge probe) ↑ higher is better · probe · secondary	0.223	[0.046, 0.364]	100	—	0.411	reproduced	unknown	Board