Benchmarks · protein-ml/protgpt2

ProteinGym protein-ml/proteingym

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

n

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	−0.156 [−0.585, 0.254]	25
MET_HUMAN_Estevam_2023	0.535 [0.106, 0.847]	25
S22A1_HUMAN_Yee_2023_activity	0.255 [−0.168, 0.653]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.409 [−0.005, 0.719]	25

0.261

[0.047, 0.445]

100

10 / 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	0.340 [0.130, 0.565]	25
MET_HUMAN_Estevam_2023	0.792 [0.569, 1.000]	25
S22A1_HUMAN_Yee_2023_activity	0.660 [0.437, 0.872]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.722 [0.492, 0.931]	25

0.629

[0.523, 0.735]

100

11 / 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	−0.282 [−0.659, 0.200]	25
MET_HUMAN_Estevam_2023	0.600 [0.025, 0.923]	25
S22A1_HUMAN_Yee_2023_activity	0.294 [−0.116, 0.690]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.447 [−0.045, 0.757]	25

0.265

[0.019, 0.451]

100

6 / 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	0.460 [0.371, 0.652]	25
MET_HUMAN_Estevam_2023	0.633 [0.419, 0.972]	25
S22A1_HUMAN_Yee_2023_activity	0.752 [0.484, 1.000]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.218 [0.079, 0.811]	25

0.516

[0.443, 0.725]

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, 0.333]	25
MET_HUMAN_Estevam_2023	0.000 [0.000, 0.667]	25
S22A1_HUMAN_Yee_2023_activity	0.000 [0.000, 0.838]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.000 [0.000, 0.750]	25

0.000

[0.000, 0.338]

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

n

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	−0.156 [−0.585, 0.254]	25
MET_HUMAN_Estevam_2023	0.535 [0.106, 0.847]	25
S22A1_HUMAN_Yee_2023_activity	0.255 [−0.168, 0.653]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.409 [−0.005, 0.719]	25

0.261

[0.047, 0.445]

100

10 / 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	0.340 [0.130, 0.565]	25
MET_HUMAN_Estevam_2023	0.792 [0.569, 1.000]	25
S22A1_HUMAN_Yee_2023_activity	0.660 [0.437, 0.872]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.722 [0.492, 0.931]	25

0.629

[0.523, 0.735]

100

11 / 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	−0.282 [−0.659, 0.200]	25
MET_HUMAN_Estevam_2023	0.600 [0.025, 0.923]	25
S22A1_HUMAN_Yee_2023_activity	0.294 [−0.116, 0.690]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.447 [−0.045, 0.757]	25

0.265

[0.019, 0.451]

100

6 / 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	0.460 [0.371, 0.652]	25
MET_HUMAN_Estevam_2023	0.633 [0.419, 0.972]	25
S22A1_HUMAN_Yee_2023_activity	0.752 [0.484, 1.000]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.218 [0.079, 0.811]	25

0.516

[0.443, 0.725]

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, 0.333]	25
MET_HUMAN_Estevam_2023	0.000 [0.000, 0.667]	25
S22A1_HUMAN_Yee_2023_activity	0.000 [0.000, 0.838]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.000 [0.000, 0.750]	25

0.000

[0.000, 0.338]

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

n

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	−0.288 [−0.648, 0.199]	25
CCR5_HUMAN_Gill_2023	−0.156 [−0.585, 0.254]	25
OPSD_HUMAN_Wan_2019	0.454 [0.043, 0.750]	25
S22A1_HUMAN_Yee_2023_activity	0.244 [−0.171, 0.651]	25

0.063

[−0.128, 0.282]

100

13 / 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	0.331 [0.110, 0.588]	25
CCR5_HUMAN_Gill_2023	0.340 [0.130, 0.565]	25
OPSD_HUMAN_Wan_2019	0.806 [0.620, 0.965]	25
S22A1_HUMAN_Yee_2023_activity	0.660 [0.442, 0.864]	25

0.534

[0.436, 0.638]

100

13 / 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	−0.206 [−0.690, 0.161]	25
CCR5_HUMAN_Gill_2023	−0.282 [−0.659, 0.200]	25
OPSD_HUMAN_Wan_2019	0.447 [−0.035, 0.850]	25
S22A1_HUMAN_Yee_2023_activity	0.294 [−0.175, 0.665]	25

0.063

[−0.158, 0.275]

100

13 / 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	0.459 [0.000, 0.962]	25
CCR5_HUMAN_Gill_2023	0.460 [0.371, 0.652]	25
OPSD_HUMAN_Wan_2019	0.415 [0.009, 0.803]	25
S22A1_HUMAN_Yee_2023_activity	0.752 [0.484, 1.000]	25

0.521

[0.345, 0.723]

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, 0.667]	25
CCR5_HUMAN_Gill_2023	0.000 [0.000, 0.333]	25
OPSD_HUMAN_Wan_2019	0.000 [0.000, 0.714]	25
S22A1_HUMAN_Yee_2023_activity	0.000 [0.000, 0.750]	25

0.000

[0.000, 0.333]

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

n

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	0.535 [0.106, 0.847]	25
MK01_HUMAN_Brenan_2016	0.308 [−0.121, 0.640]	25
SCN5A_HUMAN_Glazer_2019	−0.130 [−0.489, 0.266]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.394 [−0.038, 0.709]	25

0.277

[0.084, 0.463]

100

4 / 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	0.792 [0.569, 1.000]	25
MK01_HUMAN_Brenan_2016	0.779 [0.544, 0.970]	25
SCN5A_HUMAN_Glazer_2019	0.400 [0.180, 0.647]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.715 [0.476, 0.931]	25

0.672

[0.563, 0.784]

100

8 / 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	0.600 [0.025, 0.923]	25
MK01_HUMAN_Brenan_2016	0.316 [−0.113, 0.665]	25
SCN5A_HUMAN_Glazer_2019	−0.294 [−0.603, 0.204]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.447 [−0.045, 0.757]	25

0.267

[0.042, 0.466]

100

4 / 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	0.633 [0.419, 0.972]	25
MK01_HUMAN_Brenan_2016	0.889 [0.487, 0.960]	25
SCN5A_HUMAN_Glazer_2019	0.404 [0.272, 0.929]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.218 [0.079, 0.811]	25

0.536

[0.422, 0.735]

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, 0.667]	25
MK01_HUMAN_Brenan_2016	0.000 [0.000, 0.667]	25
SCN5A_HUMAN_Glazer_2019	0.000 [0.000, 1.000]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.000 [0.000, 0.750]	25

0.000

[0.000, 0.375]

100

—

0.250

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

n

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	0.271 [0.079, 0.453]	100
CASP3_HUMAN_Roychowdhury_2020	0.197 [0.010, 0.379]	100
CCR5_HUMAN_Gill_2023	0.228 [0.027, 0.404]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.284 [0.092, 0.461]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	−0.089 [−0.282, 0.091]	100
KCNE1_HUMAN_Muhammad_2023_expression	−0.092 [−0.270, 0.119]	100
MLAC_ECOLI_MacRae_2023	0.122 [−0.097, 0.316]	100
MTH3_HAEAE_RockahShmuel_2015	0.331 [0.120, 0.504]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.526 [0.353, 0.665]	100
SPG1_STRSG_Olson_2014	0.045 [−0.168, 0.261]	100

0.182

[0.121, 0.242]

1000

11 / 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	0.640 [0.528, 0.757]	100
CASP3_HUMAN_Roychowdhury_2020	0.587 [0.485, 0.701]	100
CCR5_HUMAN_Gill_2023	0.590 [0.469, 0.701]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.632 [0.519, 0.732]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.425 [0.305, 0.549]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.484 [0.346, 0.624]	100
MLAC_ECOLI_MacRae_2023	0.586 [0.465, 0.700]	100
MTH3_HAEAE_RockahShmuel_2015	0.734 [0.603, 0.845]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.828 [0.739, 0.904]	100
SPG1_STRSG_Olson_2014	0.561 [0.379, 0.740]	100

0.607

[0.564, 0.647]

1000

11 / 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	0.140 [−0.062, 0.343]	100
CASP3_HUMAN_Roychowdhury_2020	0.100 [−0.095, 0.301]	100
CCR5_HUMAN_Gill_2023	0.200 [−0.061, 0.360]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.161 [−0.052, 0.360]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	−0.175 [−0.368, 0.032]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.089 [−0.102, 0.277]	100
MLAC_ECOLI_MacRae_2023	0.160 [−0.058, 0.380]	100
MTH3_HAEAE_RockahShmuel_2015	0.281 [0.094, 0.445]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.490 [0.295, 0.663]	100
SPG1_STRSG_Olson_2014	0.058 [−0.167, 0.293]	100

0.150

[0.080, 0.215]

1000

11 / 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	0.718 [0.575, 0.880]	100
CASP3_HUMAN_Roychowdhury_2020	0.655 [0.563, 0.808]	100
CCR5_HUMAN_Gill_2023	0.822 [0.649, 0.867]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.594 [0.375, 0.760]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.775 [0.697, 0.868]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.380 [0.210, 0.536]	100
MLAC_ECOLI_MacRae_2023	0.851 [0.673, 0.938]	100
MTH3_HAEAE_RockahShmuel_2015	0.466 [0.205, 0.796]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.875 [0.750, 0.926]	100
SPG1_STRSG_Olson_2014	0.646 [0.458, 0.840]	100

0.678

[0.620, 0.730]

1000

11 / 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	0.000 [0.000, 0.300]	100
CASP3_HUMAN_Roychowdhury_2020	0.100 [0.000, 0.500]	100
CCR5_HUMAN_Gill_2023	0.000 [0.000, 0.273]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.200 [0.000, 0.500]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.000 [0.000, 0.273]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.000 [0.000, 0.000]	100
MLAC_ECOLI_MacRae_2023	0.200 [0.000, 0.500]	100
MTH3_HAEAE_RockahShmuel_2015	0.182 [0.000, 0.400]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.300 [0.000, 0.500]	100
SPG1_STRSG_Olson_2014	0.000 [0.000, 0.300]	100

0.098

[0.056, 0.190]

1000

11 / 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	0.513 [0.345, 0.647]	100
CASP3_HUMAN_Roychowdhury_2020	0.462 [0.289, 0.611]	100
CCR5_HUMAN_Gill_2023	0.193 [−0.016, 0.397]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.478 [0.319, 0.609]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.585 [0.409, 0.721]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.502 [0.304, 0.683]	100
MLAC_ECOLI_MacRae_2023	0.393 [0.209, 0.562]	100
MTH3_HAEAE_RockahShmuel_2015	0.584 [0.432, 0.708]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.839 [0.766, 0.889]	100
SPG1_STRSG_Olson_2014	0.767 [0.656, 0.844]	100

0.532

[0.478, 0.576]

1000

13 / 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	0.767 [0.669, 0.860]	100
CASP3_HUMAN_Roychowdhury_2020	0.748 [0.656, 0.838]	100
CCR5_HUMAN_Gill_2023	0.657 [0.544, 0.766]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.754 [0.659, 0.843]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.830 [0.723, 0.918]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.795 [0.683, 0.893]	100
MLAC_ECOLI_MacRae_2023	0.726 [0.622, 0.831]	100
MTH3_HAEAE_RockahShmuel_2015	0.766 [0.648, 0.859]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.920 [0.867, 0.971]	100
SPG1_STRSG_Olson_2014	0.940 [0.875, 0.988]	100

0.790

[0.761, 0.818]

1000

12 / 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	0.420 [0.241, 0.620]	100
CASP3_HUMAN_Roychowdhury_2020	0.341 [0.193, 0.546]	100
CCR5_HUMAN_Gill_2023	0.160 [−0.040, 0.380]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.361 [0.160, 0.540]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.436 [0.234, 0.577]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.356 [0.141, 0.552]	100
MLAC_ECOLI_MacRae_2023	0.360 [0.182, 0.580]	100
MTH3_HAEAE_RockahShmuel_2015	0.375 [0.170, 0.511]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.653 [0.471, 0.800]	100
SPG1_STRSG_Olson_2014	0.403 [0.278, 0.500]	100

0.387

[0.329, 0.441]

1000

12 / 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	0.766 [0.666, 0.900]	100
CASP3_HUMAN_Roychowdhury_2020	0.895 [0.737, 0.957]	100
CCR5_HUMAN_Gill_2023	0.817 [0.584, 0.880]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.829 [0.722, 0.890]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.831 [0.720, 0.926]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.675 [0.539, 0.848]	100
MLAC_ECOLI_MacRae_2023	0.884 [0.778, 0.948]	100
MTH3_HAEAE_RockahShmuel_2015	0.748 [0.603, 0.907]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.952 [0.904, 0.978]	100
SPG1_STRSG_Olson_2014	0.926 [0.858, 0.977]	100

0.832

[0.787, 0.862]

1000

7 / 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	0.100 [0.000, 0.357]	100
CASP3_HUMAN_Roychowdhury_2020	0.400 [0.091, 0.700]	100
CCR5_HUMAN_Gill_2023	0.200 [0.000, 0.400]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.300 [0.000, 0.500]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.300 [0.000, 0.636]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.100 [0.000, 0.500]	100
MLAC_ECOLI_MacRae_2023	0.300 [0.000, 0.500]	100
MTH3_HAEAE_RockahShmuel_2015	0.182 [0.000, 0.500]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.500 [0.200, 0.900]	100
SPG1_STRSG_Olson_2014	0.300 [0.091, 0.600]	100

0.268

[0.174, 0.352]

1000

9 / 17

0.360

reproduced

overlap

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

Board

ProteinGym_v1 protein-ml/proteingym-v1

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

n

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	−0.156 [−0.585, 0.254]	25
MET_HUMAN_Estevam_2023	0.535 [0.106, 0.847]	25
S22A1_HUMAN_Yee_2023_activity	0.255 [−0.168, 0.653]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.409 [−0.005, 0.719]	25

0.261

[0.047, 0.445]

100

10 / 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	0.340 [0.130, 0.565]	25
MET_HUMAN_Estevam_2023	0.792 [0.569, 1.000]	25
S22A1_HUMAN_Yee_2023_activity	0.660 [0.437, 0.872]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.722 [0.492, 0.931]	25

0.629

[0.523, 0.735]

100

11 / 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	−0.282 [−0.659, 0.200]	25
MET_HUMAN_Estevam_2023	0.600 [0.025, 0.923]	25
S22A1_HUMAN_Yee_2023_activity	0.294 [−0.116, 0.690]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.447 [−0.045, 0.757]	25

0.265

[0.019, 0.451]

100

6 / 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	0.460 [0.371, 0.652]	25
MET_HUMAN_Estevam_2023	0.633 [0.419, 0.972]	25
S22A1_HUMAN_Yee_2023_activity	0.752 [0.484, 1.000]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.218 [0.079, 0.811]	25

0.516

[0.443, 0.725]

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, 0.333]	25
MET_HUMAN_Estevam_2023	0.000 [0.000, 0.667]	25
S22A1_HUMAN_Yee_2023_activity	0.000 [0.000, 0.838]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.000 [0.000, 0.750]	25

0.000

[0.000, 0.338]

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

n

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	−0.156 [−0.585, 0.254]	25
MET_HUMAN_Estevam_2023	0.535 [0.106, 0.847]	25
S22A1_HUMAN_Yee_2023_activity	0.255 [−0.168, 0.653]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.409 [−0.005, 0.719]	25

0.261

[0.047, 0.445]

100

10 / 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	0.340 [0.130, 0.565]	25
MET_HUMAN_Estevam_2023	0.792 [0.569, 1.000]	25
S22A1_HUMAN_Yee_2023_activity	0.660 [0.437, 0.872]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.722 [0.492, 0.931]	25

0.629

[0.523, 0.735]

100

11 / 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	−0.282 [−0.659, 0.200]	25
MET_HUMAN_Estevam_2023	0.600 [0.025, 0.923]	25
S22A1_HUMAN_Yee_2023_activity	0.294 [−0.116, 0.690]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.447 [−0.045, 0.757]	25

0.265

[0.019, 0.451]

100

6 / 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	0.460 [0.371, 0.652]	25
MET_HUMAN_Estevam_2023	0.633 [0.419, 0.972]	25
S22A1_HUMAN_Yee_2023_activity	0.752 [0.484, 1.000]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.218 [0.079, 0.811]	25

0.516

[0.443, 0.725]

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, 0.333]	25
MET_HUMAN_Estevam_2023	0.000 [0.000, 0.667]	25
S22A1_HUMAN_Yee_2023_activity	0.000 [0.000, 0.838]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.000 [0.000, 0.750]	25

0.000

[0.000, 0.338]

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

n

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	−0.288 [−0.648, 0.199]	25
CCR5_HUMAN_Gill_2023	−0.156 [−0.585, 0.254]	25
OPSD_HUMAN_Wan_2019	0.454 [0.043, 0.750]	25
S22A1_HUMAN_Yee_2023_activity	0.244 [−0.171, 0.651]	25

0.063

[−0.128, 0.282]

100

13 / 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	0.331 [0.110, 0.588]	25
CCR5_HUMAN_Gill_2023	0.340 [0.130, 0.565]	25
OPSD_HUMAN_Wan_2019	0.806 [0.620, 0.965]	25
S22A1_HUMAN_Yee_2023_activity	0.660 [0.442, 0.864]	25

0.534

[0.436, 0.638]

100

13 / 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	−0.206 [−0.690, 0.161]	25
CCR5_HUMAN_Gill_2023	−0.282 [−0.659, 0.200]	25
OPSD_HUMAN_Wan_2019	0.447 [−0.035, 0.850]	25
S22A1_HUMAN_Yee_2023_activity	0.294 [−0.175, 0.665]	25

0.063

[−0.158, 0.275]

100

13 / 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	0.459 [0.000, 0.962]	25
CCR5_HUMAN_Gill_2023	0.460 [0.371, 0.652]	25
OPSD_HUMAN_Wan_2019	0.415 [0.009, 0.803]	25
S22A1_HUMAN_Yee_2023_activity	0.752 [0.484, 1.000]	25

0.521

[0.345, 0.723]

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, 0.667]	25
CCR5_HUMAN_Gill_2023	0.000 [0.000, 0.333]	25
OPSD_HUMAN_Wan_2019	0.000 [0.000, 0.714]	25
S22A1_HUMAN_Yee_2023_activity	0.000 [0.000, 0.750]	25

0.000

[0.000, 0.333]

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

n

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	0.535 [0.106, 0.847]	25
MK01_HUMAN_Brenan_2016	0.308 [−0.121, 0.640]	25
SCN5A_HUMAN_Glazer_2019	−0.130 [−0.489, 0.266]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.394 [−0.038, 0.709]	25

0.277

[0.084, 0.463]

100

4 / 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	0.792 [0.569, 1.000]	25
MK01_HUMAN_Brenan_2016	0.779 [0.544, 0.970]	25
SCN5A_HUMAN_Glazer_2019	0.400 [0.180, 0.647]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.715 [0.476, 0.931]	25

0.672

[0.563, 0.784]

100

8 / 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	0.600 [0.025, 0.923]	25
MK01_HUMAN_Brenan_2016	0.316 [−0.113, 0.665]	25
SCN5A_HUMAN_Glazer_2019	−0.294 [−0.603, 0.204]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.447 [−0.045, 0.757]	25

0.267

[0.042, 0.466]

100

4 / 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	0.633 [0.419, 0.972]	25
MK01_HUMAN_Brenan_2016	0.889 [0.487, 0.960]	25
SCN5A_HUMAN_Glazer_2019	0.404 [0.272, 0.929]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.218 [0.079, 0.811]	25

0.536

[0.422, 0.735]

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, 0.667]	25
MK01_HUMAN_Brenan_2016	0.000 [0.000, 0.667]	25
SCN5A_HUMAN_Glazer_2019	0.000 [0.000, 1.000]	25
SRC_HUMAN_Chakraborty_2023_binding-DAS_25uM	0.000 [0.000, 0.750]	25

0.000

[0.000, 0.375]

100

—

0.250

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

n

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	0.271 [0.079, 0.453]	100
CASP3_HUMAN_Roychowdhury_2020	0.197 [0.010, 0.379]	100
CCR5_HUMAN_Gill_2023	0.228 [0.027, 0.404]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.284 [0.092, 0.461]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	−0.089 [−0.282, 0.091]	100
KCNE1_HUMAN_Muhammad_2023_expression	−0.092 [−0.270, 0.119]	100
MLAC_ECOLI_MacRae_2023	0.122 [−0.097, 0.316]	100
MTH3_HAEAE_RockahShmuel_2015	0.331 [0.120, 0.504]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.526 [0.353, 0.665]	100
SPG1_STRSG_Olson_2014	0.045 [−0.168, 0.261]	100

0.182

[0.121, 0.242]

1000

11 / 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	0.640 [0.528, 0.757]	100
CASP3_HUMAN_Roychowdhury_2020	0.587 [0.485, 0.701]	100
CCR5_HUMAN_Gill_2023	0.590 [0.469, 0.701]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.632 [0.519, 0.732]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.425 [0.305, 0.549]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.484 [0.346, 0.624]	100
MLAC_ECOLI_MacRae_2023	0.586 [0.465, 0.700]	100
MTH3_HAEAE_RockahShmuel_2015	0.734 [0.603, 0.845]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.828 [0.739, 0.904]	100
SPG1_STRSG_Olson_2014	0.561 [0.379, 0.740]	100

0.607

[0.564, 0.647]

1000

11 / 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	0.140 [−0.062, 0.343]	100
CASP3_HUMAN_Roychowdhury_2020	0.100 [−0.095, 0.301]	100
CCR5_HUMAN_Gill_2023	0.200 [−0.061, 0.360]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.161 [−0.052, 0.360]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	−0.175 [−0.368, 0.032]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.089 [−0.102, 0.277]	100
MLAC_ECOLI_MacRae_2023	0.160 [−0.058, 0.380]	100
MTH3_HAEAE_RockahShmuel_2015	0.281 [0.094, 0.445]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.490 [0.295, 0.663]	100
SPG1_STRSG_Olson_2014	0.058 [−0.167, 0.293]	100

0.150

[0.080, 0.215]

1000

11 / 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	0.718 [0.575, 0.880]	100
CASP3_HUMAN_Roychowdhury_2020	0.655 [0.563, 0.808]	100
CCR5_HUMAN_Gill_2023	0.822 [0.649, 0.867]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.594 [0.375, 0.760]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.775 [0.697, 0.868]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.380 [0.210, 0.536]	100
MLAC_ECOLI_MacRae_2023	0.851 [0.673, 0.938]	100
MTH3_HAEAE_RockahShmuel_2015	0.466 [0.205, 0.796]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.875 [0.750, 0.926]	100
SPG1_STRSG_Olson_2014	0.646 [0.458, 0.840]	100

0.678

[0.620, 0.730]

1000

11 / 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	0.000 [0.000, 0.300]	100
CASP3_HUMAN_Roychowdhury_2020	0.100 [0.000, 0.500]	100
CCR5_HUMAN_Gill_2023	0.000 [0.000, 0.273]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.200 [0.000, 0.500]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.000 [0.000, 0.273]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.000 [0.000, 0.000]	100
MLAC_ECOLI_MacRae_2023	0.200 [0.000, 0.500]	100
MTH3_HAEAE_RockahShmuel_2015	0.182 [0.000, 0.400]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.300 [0.000, 0.500]	100
SPG1_STRSG_Olson_2014	0.000 [0.000, 0.300]	100

0.098

[0.056, 0.190]

1000

11 / 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	0.513 [0.345, 0.647]	100
CASP3_HUMAN_Roychowdhury_2020	0.462 [0.289, 0.611]	100
CCR5_HUMAN_Gill_2023	0.193 [−0.016, 0.397]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.478 [0.319, 0.609]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.585 [0.409, 0.721]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.502 [0.304, 0.683]	100
MLAC_ECOLI_MacRae_2023	0.393 [0.209, 0.562]	100
MTH3_HAEAE_RockahShmuel_2015	0.584 [0.432, 0.708]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.839 [0.766, 0.889]	100
SPG1_STRSG_Olson_2014	0.767 [0.656, 0.844]	100

0.532

[0.478, 0.576]

1000

13 / 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	0.767 [0.669, 0.860]	100
CASP3_HUMAN_Roychowdhury_2020	0.748 [0.656, 0.838]	100
CCR5_HUMAN_Gill_2023	0.657 [0.544, 0.766]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.754 [0.659, 0.843]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.830 [0.723, 0.918]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.795 [0.683, 0.893]	100
MLAC_ECOLI_MacRae_2023	0.726 [0.622, 0.831]	100
MTH3_HAEAE_RockahShmuel_2015	0.766 [0.648, 0.859]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.920 [0.867, 0.971]	100
SPG1_STRSG_Olson_2014	0.940 [0.875, 0.988]	100

0.790

[0.761, 0.818]

1000

12 / 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	0.420 [0.241, 0.620]	100
CASP3_HUMAN_Roychowdhury_2020	0.341 [0.193, 0.546]	100
CCR5_HUMAN_Gill_2023	0.160 [−0.040, 0.380]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.361 [0.160, 0.540]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.436 [0.234, 0.577]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.356 [0.141, 0.552]	100
MLAC_ECOLI_MacRae_2023	0.360 [0.182, 0.580]	100
MTH3_HAEAE_RockahShmuel_2015	0.375 [0.170, 0.511]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.653 [0.471, 0.800]	100
SPG1_STRSG_Olson_2014	0.403 [0.278, 0.500]	100

0.387

[0.329, 0.441]

1000

12 / 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	0.766 [0.666, 0.900]	100
CASP3_HUMAN_Roychowdhury_2020	0.895 [0.737, 0.957]	100
CCR5_HUMAN_Gill_2023	0.817 [0.584, 0.880]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.829 [0.722, 0.890]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.831 [0.720, 0.926]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.675 [0.539, 0.848]	100
MLAC_ECOLI_MacRae_2023	0.884 [0.778, 0.948]	100
MTH3_HAEAE_RockahShmuel_2015	0.748 [0.603, 0.907]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.952 [0.904, 0.978]	100
SPG1_STRSG_Olson_2014	0.926 [0.858, 0.977]	100

0.832

[0.787, 0.862]

1000

7 / 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	0.100 [0.000, 0.357]	100
CASP3_HUMAN_Roychowdhury_2020	0.400 [0.091, 0.700]	100
CCR5_HUMAN_Gill_2023	0.200 [0.000, 0.400]	100
CP2C9_HUMAN_Amorosi_2021_abundance	0.300 [0.000, 0.500]	100
ILF3_HUMAN_Tsuboyama_2023_2L33	0.300 [0.000, 0.636]	100
KCNE1_HUMAN_Muhammad_2023_expression	0.100 [0.000, 0.500]	100
MLAC_ECOLI_MacRae_2023	0.300 [0.000, 0.500]	100
MTH3_HAEAE_RockahShmuel_2015	0.182 [0.000, 0.500]	100
RPC1_BP434_Tsuboyama_2023_1R69	0.500 [0.200, 0.900]	100
SPG1_STRSG_Olson_2014	0.300 [0.091, 0.600]	100

0.268

[0.174, 0.352]

1000

9 / 17

0.360

reproduced

overlap

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

Board

tape protein-ml/tape

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	0.616	[0.600, 0.630]	100	17 / 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	0.460	[0.441, 0.477]	100	17 / 17	0.663	reproduced	overlap Train/test overlap declared by the evaluator: protein-ml/uniref50	Board

UniProt Knowledgebase protein-ml/uniprot-knowledgebase

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	0.707	[0.653, 0.775]	100	4 / 17	0.844	reproduced	unknown	Board
Fmax (biological process, probe) ↑ higher is better · probe	0.412	[0.323, 0.536]	100	17 / 17	0.538	reproduced	unknown	Board
Fmax (cellular component, probe) ↑ higher is better · probe	0.728	[0.675, 0.791]	100	9 / 17	0.756	reproduced	unknown	Board
Micro AUPR (molecular function, probe) ↑ higher is better · probe · secondary	0.706	[0.618, 0.801]	100	—	0.875	reproduced	unknown	Board
Smin (molecular function, probe) ↓ lower is better · bits · probe · secondary	2.887	[2.157, 3.601]	100	—	1.727	reproduced	unknown	Board
Micro AUPR (biological process, probe) ↑ higher is better · probe · secondary	0.428	[0.308, 0.535]	100	—	0.595	reproduced	unknown	Board
Smin (biological process, probe) ↓ lower is better · bits · probe · secondary	6.429	[4.569, 7.880]	100	—	4.842	reproduced	unknown	Board
Micro AUPR (cellular component, probe) ↑ higher is better · probe · secondary	0.707	[0.639, 0.772]	100	—	0.758	reproduced	unknown	Board
Smin (cellular component, probe) ↓ lower is better · bits · probe · secondary	3.504	[2.705, 4.227]	100	—	3.092	reproduced	unknown	Board

Other Benchmarks drawn from no hub dataset you can see

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	0.822	[0.805, 0.840]	1728	15 / 17	0.888	reproduced	overlap Train/test overlap declared by the evaluator: protein-ml/uniref50	Board
MCC (logistic probe) ↑ higher is better · probe · secondary	0.635	[0.600, 0.672]	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	0.345	[0.202, 0.470]	100	3 / 12	0.410	reproduced	unknown	Board
Spearman ρ (ridge probe) ↑ higher is better · probe	0.160	[−0.062, 0.317]	100	7 / 17	0.297	reproduced	unknown	Board
RMSE, kcal/mol (ridge probe) ↓ lower is better · kcal/mol · probe	1.202	[0.990, 1.386]	100	16 / 17	1.095	reproduced	unknown	Board
Pearson r (ridge probe) ↑ higher is better · probe · secondary	0.219	[0.004, 0.370]	100	—	0.349	reproduced	unknown	Board
MAE, kcal/mol (ridge probe) ↓ lower is better · kcal/mol · probe · secondary	0.968	[0.769, 1.164]	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	0.677	[0.662, 0.691]	2742	9 / 17	0.801	reproduced	overlap Train/test overlap declared by the evaluator: protein-ml/uniref50	Board
Macro F1 (logistic probe) ↑ higher is better · probe	0.481	[0.464, 0.498]	2742	8 / 17	0.640	reproduced	overlap Train/test overlap declared by the evaluator: protein-ml/uniref50	Board
MCC (logistic probe) ↑ higher is better · probe · secondary	0.605	[0.587, 0.624]	2742	—	0.758	reproduced	overlap Train/test overlap declared by the evaluator: protein-ml/uniref50	Board