Models / protein-ml / esm2-t12-35m-ur50d / benchmarks

protein-ml/esm2-t12-35m-ur50d

Visibility: PUBLIC

Model card FilesBenchmarks

Group by

Domain variant fitness, human PLM 0.5 (ProteinGym x Pfam) protein-ml/protein-domain-variant-fitness-pfam-05

Drawn from: ProteinGym, ProteinGym_v1

atlas-n25-v2

Metric

Value

95% CI

Rank

Best in catalogue

Source

Overlap

Spearman ρ (zero-shot, domain input) ↑ higher is better · zero-shot

By assay: 4 levels

assay	Value [95% CI]	n
CCR5_HUMAN_Gill_2023	0.312 [−0.097, 0.677]	25
MET_HUMAN_Estevam_2023	0.292 [−0.119, 0.644]	25
S22A1_HUMAN_Yee_2023_activity	0.465 [0.049, 0.773]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.060 [−0.294, 0.453]	25

0.282

[0.062, 0.462]

100

8 / 13

0.476

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

AUC (zero-shot, domain input) ↑ higher is better · zero-shot

By assay: 4 levels

assay	Value [95% CI]	n
CCR5_HUMAN_Gill_2023	0.667 [0.448, 0.870]	25
MET_HUMAN_Estevam_2023	0.675 [0.449, 0.907]	25
S22A1_HUMAN_Yee_2023_activity	0.693 [0.419, 0.903]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.590 [0.353, 0.819]	25

0.656

[0.536, 0.764]

100

9 / 13

0.759

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

MCC (zero-shot, domain input) ↑ higher is better · zero-shot

By assay: 4 levels

assay	Value [95% CI]	n
CCR5_HUMAN_Gill_2023	0.359 [−0.137, 0.621]	25
MET_HUMAN_Estevam_2023	0.277 [−0.242, 0.761]	25
S22A1_HUMAN_Yee_2023_activity	0.131 [−0.199, 0.612]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.113 [−0.373, 0.600]	25

0.220

[−0.013, 0.450]

100

9 / 13

0.426

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

NDCG, top 10 % (zero-shot, domain input) ↑ higher is better · zero-shot · secondary

By assay: 4 levels

assay	Value [95% CI]	n
CCR5_HUMAN_Gill_2023	0.300 [0.000, 0.718]	25
MET_HUMAN_Estevam_2023	0.597 [0.391, 0.945]	25
S22A1_HUMAN_Yee_2023_activity	0.465 [0.089, 0.939]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.109 [0.028, 0.237]	25

0.368

[0.247, 0.577]

100

—

0.635

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

Top-10 % recall (zero-shot, domain input) ↑ higher is better · zero-shot · secondary

By assay: 4 levels

assay	Value [95% CI]	n
CCR5_HUMAN_Gill_2023	0.000 [0.000, 0.005]	25
MET_HUMAN_Estevam_2023	0.000 [0.000, 0.333]	25
S22A1_HUMAN_Yee_2023_activity	0.000 [0.000, 1.000]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.000 [0.000, 0.000]	25

0.000

[0.000, 0.250]

100

—

0.167

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

Domain variant fitness, human PLM 0.9 (ProteinGym x Pfam) protein-ml/protein-domain-variant-fitness-pfam-09

Drawn from: ProteinGym, ProteinGym_v1

atlas-n25-v2

Metric

Value

95% CI

Rank

Best in catalogue

Source

Overlap

Spearman ρ (zero-shot, domain input) ↑ higher is better · zero-shot

By assay: 4 levels

assay	Value [95% CI]	n
CCR5_HUMAN_Gill_2023	0.312 [−0.097, 0.677]	25
MET_HUMAN_Estevam_2023	0.292 [−0.119, 0.644]	25
S22A1_HUMAN_Yee_2023_activity	0.465 [0.049, 0.773]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.060 [−0.294, 0.453]	25

0.282

[0.062, 0.462]

100

8 / 13

0.476

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

AUC (zero-shot, domain input) ↑ higher is better · zero-shot

By assay: 4 levels

assay	Value [95% CI]	n
CCR5_HUMAN_Gill_2023	0.667 [0.448, 0.870]	25
MET_HUMAN_Estevam_2023	0.675 [0.449, 0.907]	25
S22A1_HUMAN_Yee_2023_activity	0.693 [0.419, 0.903]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.590 [0.353, 0.819]	25

0.656

[0.536, 0.764]

100

9 / 13

0.759

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

MCC (zero-shot, domain input) ↑ higher is better · zero-shot

By assay: 4 levels

assay	Value [95% CI]	n
CCR5_HUMAN_Gill_2023	0.359 [−0.137, 0.621]	25
MET_HUMAN_Estevam_2023	0.277 [−0.242, 0.761]	25
S22A1_HUMAN_Yee_2023_activity	0.131 [−0.199, 0.612]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.113 [−0.373, 0.600]	25

0.220

[−0.013, 0.450]

100

9 / 13

0.426

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

NDCG, top 10 % (zero-shot, domain input) ↑ higher is better · zero-shot · secondary

By assay: 4 levels

assay	Value [95% CI]	n
CCR5_HUMAN_Gill_2023	0.300 [0.000, 0.718]	25
MET_HUMAN_Estevam_2023	0.597 [0.391, 0.945]	25
S22A1_HUMAN_Yee_2023_activity	0.465 [0.089, 0.939]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.109 [0.028, 0.237]	25

0.368

[0.247, 0.577]

100

—

0.635

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

Top-10 % recall (zero-shot, domain input) ↑ higher is better · zero-shot · secondary

By assay: 4 levels

assay	Value [95% CI]	n
CCR5_HUMAN_Gill_2023	0.000 [0.000, 0.005]	25
MET_HUMAN_Estevam_2023	0.000 [0.000, 0.333]	25
S22A1_HUMAN_Yee_2023_activity	0.000 [0.000, 1.000]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.000 [0.000, 0.000]	25

0.000

[0.000, 0.250]

100

—

0.167

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

Domain variant fitness, human PLM Clans-0.5 (ProteinGym x Pfam) protein-ml/protein-domain-variant-fitness-clan-05

Drawn from: ProteinGym, ProteinGym_v1

atlas-n25-v2

Metric

Value

95% CI

Rank

Best in catalogue

Source

Overlap

Spearman ρ (zero-shot, domain input) ↑ higher is better · zero-shot

By assay: 4 levels

assay	Value [95% CI]	n
ADRB2_HUMAN_Jones_2020	0.043 [−0.378, 0.435]	25
CCR5_HUMAN_Gill_2023	0.312 [−0.097, 0.677]	25
OPSD_HUMAN_Wan_2019	0.313 [−0.084, 0.675]	25
S22A1_HUMAN_Yee_2023_activity	0.462 [0.041, 0.768]	25

0.282

[0.090, 0.488]

100

10 / 13

0.463

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

AUC (zero-shot, domain input) ↑ higher is better · zero-shot

By assay: 4 levels

assay	Value [95% CI]	n
ADRB2_HUMAN_Jones_2020	0.500 [0.250, 0.756]	25
CCR5_HUMAN_Gill_2023	0.667 [0.448, 0.870]	25
OPSD_HUMAN_Wan_2019	0.674 [0.449, 0.880]	25
S22A1_HUMAN_Yee_2023_activity	0.693 [0.419, 0.903]	25

0.633

[0.519, 0.748]

100

10 / 13

0.751

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

MCC (zero-shot, domain input) ↑ higher is better · zero-shot

By assay: 4 levels

assay	Value [95% CI]	n
ADRB2_HUMAN_Jones_2020	−0.045 [−0.387, 0.513]	25
CCR5_HUMAN_Gill_2023	0.359 [−0.137, 0.621]	25
OPSD_HUMAN_Wan_2019	0.113 [−0.207, 0.603]	25
S22A1_HUMAN_Yee_2023_activity	0.131 [−0.199, 0.608]	25

0.139

[−0.011, 0.417]

100

9 / 13

0.426

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

NDCG, top 10 % (zero-shot, domain input) ↑ higher is better · zero-shot · secondary

By assay: 4 levels

assay	Value [95% CI]	n
ADRB2_HUMAN_Jones_2020	0.160 [0.000, 0.750]	25
CCR5_HUMAN_Gill_2023	0.300 [0.000, 0.718]	25
OPSD_HUMAN_Wan_2019	0.253 [0.000, 0.779]	25
S22A1_HUMAN_Yee_2023_activity	0.465 [0.089, 0.939]	25

0.295

[0.145, 0.601]

100

—

0.709

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

Top-10 % recall (zero-shot, domain input) ↑ higher is better · zero-shot · secondary

By assay: 4 levels

assay	Value [95% CI]	n
ADRB2_HUMAN_Jones_2020	0.000 [0.000, 0.750]	25
CCR5_HUMAN_Gill_2023	0.000 [0.000, 0.005]	25
OPSD_HUMAN_Wan_2019	0.000 [0.000, 0.750]	25
S22A1_HUMAN_Yee_2023_activity	0.000 [0.000, 1.000]	25

0.000

[0.000, 0.362]

100

—

0.250

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

Domain variant fitness, human PLM Clans-0.9 (ProteinGym x Pfam) protein-ml/protein-domain-variant-fitness-clan-09

Drawn from: ProteinGym, ProteinGym_v1

atlas-n25-v2

Metric

Value

95% CI

Rank

Best in catalogue

Source

Overlap

Spearman ρ (zero-shot, domain input) ↑ higher is better · zero-shot

By assay: 4 levels

assay	Value [95% CI]	n
MET_HUMAN_Estevam_2023	0.293 [−0.114, 0.645]	25
MK01_HUMAN_Brenan_2016	0.397 [−0.070, 0.770]	25
SCN5A_HUMAN_Glazer_2019	−0.077 [−0.510, 0.355]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.060 [−0.294, 0.453]	25

0.168

[−0.055, 0.362]

100

11 / 13

0.362

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

AUC (zero-shot, domain input) ↑ higher is better · zero-shot

By assay: 4 levels

assay	Value [95% CI]	n
MET_HUMAN_Estevam_2023	0.675 [0.449, 0.907]	25
MK01_HUMAN_Brenan_2016	0.757 [0.463, 1.000]	25
SCN5A_HUMAN_Glazer_2019	0.407 [0.167, 0.635]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.590 [0.353, 0.819]	25

0.607

[0.485, 0.734]

100

11 / 13

0.722

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

MCC (zero-shot, domain input) ↑ higher is better · zero-shot

By assay: 4 levels

assay	Value [95% CI]	n
MET_HUMAN_Estevam_2023	0.277 [−0.242, 0.761]	25
MK01_HUMAN_Brenan_2016	0.316 [−0.115, 0.659]	25
SCN5A_HUMAN_Glazer_2019	−0.131 [−0.529, 0.263]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.113 [−0.373, 0.600]	25

0.144

[−0.101, 0.343]

100

12 / 13

0.310

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

NDCG, top 10 % (zero-shot, domain input) ↑ higher is better · zero-shot · secondary

By assay: 4 levels

assay	Value [95% CI]	n
MET_HUMAN_Estevam_2023	0.597 [0.391, 0.952]	25
MK01_HUMAN_Brenan_2016	0.399 [0.000, 1.000]	25
SCN5A_HUMAN_Glazer_2019	0.246 [0.069, 0.944]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.109 [0.028, 0.237]	25

0.338

[0.208, 0.594]

100

—

0.725

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

Top-10 % recall (zero-shot, domain input) ↑ higher is better · zero-shot · secondary

By assay: 4 levels

assay	Value [95% CI]	n
MET_HUMAN_Estevam_2023	0.000 [0.000, 0.500]	25
MK01_HUMAN_Brenan_2016	0.333 [0.000, 1.000]	25
SCN5A_HUMAN_Glazer_2019	0.000 [0.000, 0.800]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.000 [0.000, 0.000]	25

0.083

[0.000, 0.350]

100

—

0.250

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

GO function prediction (Swiss-Prot 2025H2) protein-ml/protein-function-go-prediction

Drawn from: UniProt Knowledgebase

atlas-n100-v1

Metric	Value	95% CI	n	Rank	Best in catalogue	Source	Overlap	Board
Fmax (molecular function, probe) ↑ higher is better · probe	0.665	[0.605, 0.746]	100	9 / 17	0.844	reproduced	unknown	Board
Fmax (biological process, probe) ↑ higher is better · probe	0.497	[0.391, 0.599]	100	4 / 17	0.538	reproduced	unknown	Board
Fmax (cellular component, probe) ↑ higher is better · probe	0.718	[0.663, 0.780]	100	15 / 17	0.756	reproduced	unknown	Board
Micro AUPR (molecular function, probe) ↑ higher is better · probe · secondary	0.670	[0.594, 0.758]	100	—	0.875	reproduced	unknown	Board
Smin (molecular function, probe) ↓ lower is better · bits · probe · secondary	3.289	[2.491, 4.058]	100	—	1.727	reproduced	unknown	Board
Micro AUPR (biological process, probe) ↑ higher is better · probe · secondary	0.430	[0.297, 0.567]	100	—	0.595	reproduced	unknown	Board
Smin (biological process, probe) ↓ lower is better · bits · probe · secondary	5.760	[4.304, 7.021]	100	—	4.842	reproduced	unknown	Board
Micro AUPR (cellular component, probe) ↑ higher is better · probe · secondary	0.688	[0.617, 0.757]	100	—	0.758	reproduced	unknown	Board
Smin (cellular component, probe) ↓ lower is better · bits · probe · secondary	3.481	[2.665, 4.208]	100	—	3.092	reproduced	unknown	Board

Protein membrane association (DeepLoc) protein-ml/protein-membrane-localization

atlas-full-v1

Metric	Value	95% CI	n	Rank	Best in catalogue	Source	Overlap	Board
Q2 accuracy (logistic probe) ↑ higher is better · probe	0.844	[0.828, 0.863]	1728	9 / 17	0.888	reproduced	overlap Train/test overlap declared by the evaluator: protein-ml/uniref50	Board
MCC (logistic probe) ↑ higher is better · probe · secondary	0.680	[0.646, 0.718]	1728	—	0.771	reproduced	overlap Train/test overlap declared by the evaluator: protein-ml/uniref50	Board

Protein stability ΔΔG (Mega-scale) protein-ml/protein-stability-ddg

atlas-n100-v1

Metric	Value	95% CI	n	Rank	Best in catalogue	Source	Overlap	Board
Spearman ρ (zero-shot) ↑ higher is better · zero-shot	0.278	[0.046, 0.472]	100	8 / 12	0.410	reproduced	unknown	Board
Spearman ρ (ridge probe) ↑ higher is better · probe	0.155	[−0.039, 0.340]	100	8 / 17	0.297	reproduced	unknown	Board
RMSE, kcal/mol (ridge probe) ↓ lower is better · kcal/mol · probe	1.103	[0.882, 1.302]	100	6 / 17	1.095	reproduced	unknown	Board
Pearson r (ridge probe) ↑ higher is better · probe · secondary	0.190	[0.012, 0.329]	100	—	0.349	reproduced	unknown	Board
MAE, kcal/mol (ridge probe) ↓ lower is better · kcal/mol · probe · secondary	0.835	[0.697, 0.991]	100	—	0.819	reproduced	unknown	Board

Protein subcellular localization (DeepLoc) protein-ml/protein-subcellular-localization

atlas-full-v1

Metric	Value	95% CI	n	Rank	Best in catalogue	Source	Overlap	Board
Accuracy (logistic probe) ↑ higher is better · probe	0.655	[0.639, 0.668]	2742	12 / 17	0.801	reproduced	overlap Train/test overlap declared by the evaluator: protein-ml/uniref50	Board
Macro F1 (logistic probe) ↑ higher is better · probe	0.463	[0.445, 0.480]	2742	13 / 17	0.640	reproduced	overlap Train/test overlap declared by the evaluator: protein-ml/uniref50	Board
MCC (logistic probe) ↑ higher is better · probe · secondary	0.579	[0.560, 0.596]	2742	—	0.758	reproduced	overlap Train/test overlap declared by the evaluator: protein-ml/uniref50	Board

ProteinGym DMS zero-shot (Spearman) protein-ml/proteingym-dms-zero-shot-spearman

Drawn from: ProteinGym, ProteinGym_v1

atlas-n100-v1

Metric

Value

95% CI

Rank

Best in catalogue

Source

Overlap

Spearman ρ (zero-shot) ↑ higher is better · zero-shot

By assay: 10 levels

assay	Value [95% CI]	n
ANCSZ_Hobbs_2022	0.495 [0.328, 0.647]	100
CASP3_HUMAN_Roychowdhury_2020	0.547 [0.394, 0.671]	100
CCR5_HUMAN_Gill_2023	0.293 [0.106, 0.494]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.645 [0.496, 0.758]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.378 [0.179, 0.565]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.048 [−0.157, 0.244]	100
MLAC_ECOLI_MacRae_2023	0.202 [0.012, 0.378]	100
MTH3_HAEAE_RockahShmuel_2015	0.396 [0.210, 0.566]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.731 [0.597, 0.825]	100
SPG1_STRSG_Olson_2014	0.139 [−0.068, 0.347]	100

0.388

[0.326, 0.443]

1000

9 / 11

0.434

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

AUC (zero-shot) ↑ higher is better · zero-shot

By assay: 10 levels

assay	Value [95% CI]	n
ANCSZ_Hobbs_2022	0.712 [0.607, 0.815]	100
CASP3_HUMAN_Roychowdhury_2020	0.783 [0.689, 0.876]	100
CCR5_HUMAN_Gill_2023	0.679 [0.574, 0.781]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.809 [0.714, 0.883]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.757 [0.647, 0.857]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.566 [0.440, 0.691]	100
MLAC_ECOLI_MacRae_2023	0.666 [0.559, 0.770]	100
MTH3_HAEAE_RockahShmuel_2015	0.779 [0.634, 0.894]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.940 [0.885, 0.981]	100
SPG1_STRSG_Olson_2014	0.779 [0.649, 0.893]	100

0.747

[0.712, 0.777]

1000

8 / 11

0.775

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

MCC (zero-shot) ↑ higher is better · zero-shot

By assay: 10 levels

assay	Value [95% CI]	n
ANCSZ_Hobbs_2022	0.300 [0.121, 0.520]	100
CASP3_HUMAN_Roychowdhury_2020	0.381 [0.180, 0.581]	100
CCR5_HUMAN_Gill_2023	0.280 [0.060, 0.441]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.482 [0.260, 0.626]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.349 [0.172, 0.532]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.045 [−0.144, 0.274]	100
MLAC_ECOLI_MacRae_2023	0.240 [0.080, 0.460]	100
MTH3_HAEAE_RockahShmuel_2015	0.281 [0.113, 0.473]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.694 [0.559, 0.862]	100
SPG1_STRSG_Olson_2014	0.288 [0.113, 0.437]	100

0.334

[0.277, 0.394]

1000

8 / 11

0.370

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

NDCG, top 10 % (zero-shot) ↑ higher is better · zero-shot

By assay: 10 levels

assay	Value [95% CI]	n
ANCSZ_Hobbs_2022	0.766 [0.669, 0.906]	100
CASP3_HUMAN_Roychowdhury_2020	0.777 [0.715, 0.881]	100
CCR5_HUMAN_Gill_2023	0.818 [0.636, 0.873]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.861 [0.718, 0.943]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.829 [0.748, 0.909]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.565 [0.406, 0.710]	100
MLAC_ECOLI_MacRae_2023	0.810 [0.725, 0.881]	100
MTH3_HAEAE_RockahShmuel_2015	0.514 [0.253, 0.712]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.845 [0.741, 0.932]	100
SPG1_STRSG_Olson_2014	0.845 [0.720, 0.942]	100

0.763

[0.712, 0.798]

1000

9 / 11

0.814

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

Top-10 % recall (zero-shot) ↑ higher is better · zero-shot

By assay: 10 levels

assay	Value [95% CI]	n
ANCSZ_Hobbs_2022	0.100 [0.000, 0.500]	100
CASP3_HUMAN_Roychowdhury_2020	0.200 [0.000, 0.500]	100
CCR5_HUMAN_Gill_2023	0.100 [0.000, 0.400]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.400 [0.100, 0.700]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.000 [0.000, 0.200]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.200 [0.000, 0.463]	100
MLAC_ECOLI_MacRae_2023	0.000 [0.000, 0.200]	100
MTH3_HAEAE_RockahShmuel_2015	0.091 [0.000, 0.357]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.300 [0.083, 0.600]	100
SPG1_STRSG_Olson_2014	0.300 [0.000, 0.500]	100

0.169

[0.100, 0.261]

1000

5 / 11

0.196

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

Spearman ρ (ridge probe) ↑ higher is better · probe

By assay: 10 levels

assay	Value [95% CI]	n
ANCSZ_Hobbs_2022	0.631 [0.492, 0.735]	100
CASP3_HUMAN_Roychowdhury_2020	0.445 [0.265, 0.598]	100
CCR5_HUMAN_Gill_2023	0.269 [0.060, 0.470]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.614 [0.456, 0.740]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.728 [0.582, 0.840]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.526 [0.340, 0.697]	100
MLAC_ECOLI_MacRae_2023	0.462 [0.271, 0.623]	100
MTH3_HAEAE_RockahShmuel_2015	0.637 [0.500, 0.747]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.883 [0.828, 0.917]	100
SPG1_STRSG_Olson_2014	0.799 [0.698, 0.869]	100

0.599

[0.546, 0.640]

1000

8 / 17

0.661

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

AUC (ridge probe) ↑ higher is better · probe

By assay: 10 levels

assay	Value [95% CI]	n
ANCSZ_Hobbs_2022	0.830 [0.748, 0.905]	100
CASP3_HUMAN_Roychowdhury_2020	0.754 [0.650, 0.848]	100
CCR5_HUMAN_Gill_2023	0.674 [0.565, 0.778]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.810 [0.714, 0.892]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.929 [0.852, 0.985]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.799 [0.682, 0.903]	100
MLAC_ECOLI_MacRae_2023	0.725 [0.613, 0.819]	100
MTH3_HAEAE_RockahShmuel_2015	0.800 [0.706, 0.888]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.968 [0.927, 0.996]	100
SPG1_STRSG_Olson_2014	0.939 [0.880, 0.979]	100

0.823

[0.795, 0.849]

1000

6 / 17

0.848

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

MCC (ridge probe) ↑ higher is better · probe

By assay: 10 levels

assay	Value [95% CI]	n
ANCSZ_Hobbs_2022	0.540 [0.302, 0.680]	100
CASP3_HUMAN_Roychowdhury_2020	0.421 [0.149, 0.613]	100
CCR5_HUMAN_Gill_2023	0.240 [0.056, 0.422]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.482 [0.258, 0.641]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.567 [0.400, 0.719]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.401 [0.223, 0.584]	100
MLAC_ECOLI_MacRae_2023	0.320 [0.121, 0.524]	100
MTH3_HAEAE_RockahShmuel_2015	0.328 [0.178, 0.520]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.776 [0.624, 0.902]	100
SPG1_STRSG_Olson_2014	0.403 [0.289, 0.521]	100

0.448

[0.384, 0.500]

1000

6 / 17

0.472

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

NDCG, top 10 % (ridge probe) ↑ higher is better · probe

By assay: 10 levels

assay	Value [95% CI]	n
ANCSZ_Hobbs_2022	0.798 [0.690, 0.943]	100
CASP3_HUMAN_Roychowdhury_2020	0.633 [0.424, 0.841]	100
CCR5_HUMAN_Gill_2023	0.742 [0.562, 0.819]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.793 [0.666, 0.875]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.929 [0.844, 0.975]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.600 [0.413, 0.809]	100
MLAC_ECOLI_MacRae_2023	0.820 [0.721, 0.894]	100
MTH3_HAEAE_RockahShmuel_2015	0.841 [0.762, 0.907]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.916 [0.863, 0.959]	100
SPG1_STRSG_Olson_2014	0.949 [0.926, 0.979]	100

0.802

[0.765, 0.839]

1000

10 / 17

0.860

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

Top-10 % recall (ridge probe) ↑ higher is better · probe

By assay: 10 levels

assay	Value [95% CI]	n
ANCSZ_Hobbs_2022	0.300 [0.000, 0.600]	100
CASP3_HUMAN_Roychowdhury_2020	0.200 [0.000, 0.455]	100
CCR5_HUMAN_Gill_2023	0.100 [0.000, 0.400]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.100 [0.000, 0.455]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.400 [0.091, 0.700]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.200 [0.000, 0.500]	100
MLAC_ECOLI_MacRae_2023	0.100 [0.000, 0.364]	100
MTH3_HAEAE_RockahShmuel_2015	0.182 [0.000, 0.545]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.500 [0.100, 0.800]	100
SPG1_STRSG_Olson_2014	0.500 [0.100, 0.727]	100

0.258

[0.166, 0.352]

1000

10 / 17

0.360

reproduced

overlap

Train/test overlap declared by the evaluator: protein-ml/uniref50

Board

Secondary structure (Q3) protein-ml/secondary-structure-q3

Drawn from: tape

atlas-n100-v2

Metric	Value	95% CI	n	Rank	Best in catalogue	Source	Overlap	Board
Q3 accuracy (per-residue probe) ↑ higher is better · probe	0.773	[0.758, 0.789]	100	6 / 17	0.813	reproduced	overlap Train/test overlap declared by the evaluator: protein-ml/uniref50	Board
Q8 accuracy (per-residue probe) ↑ higher is better · probe	0.612	[0.591, 0.633]	100	6 / 17	0.663	reproduced	overlap Train/test overlap declared by the evaluator: protein-ml/uniref50	Board