datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
EC_AlphaFold2
EC Dataset with AlphaFold2 Structural Sequence
Description: The Enzyme Commission number (EC number) is a numerical classification scheme for enzymes, based on the chemical reactions they catalyze.
Number of labels: 585
Problem Type: multi_label_classification
Columns:
aa_seq: protein amino acid sequence
foldseek_seq: foldseek 20 3di structural sequence
ss8_seq: DSSP 8 secondary structure sequence
Github
Simple, Efficient and Scalable Structure-aware Adapter… See the full description on the dataset page: https://huggingface.co/datasets/AI4Protein/EC_AlphaFold2.MetalIonBinding_AlphaFold2
MetalIonBinding Dataset with AlphaFold2 Structural Sequence
Description: Metal-binding proteins are proteins or protein domains that chelate a metal ion.
Number of labels: 2
Problem Type: single_label_classification
Columns:
aa_seq: protein amino acid sequence
foldseek_seq: foldseek 20 3di structural sequence
ss8_seq: DSSP 8 secondary structure sequence
ss3_seq: DSSP 3 secondary structure sequence
esm3_structure_seq: ESM3 structure sequence encoded by VQ-VAE
Github… See the full description on the dataset page: https://huggingface.co/datasets/AI4Protein/MetalIonBinding_AlphaFold2.Thermostability_AlphaFold2
Thermostability Dataset with AlphaFold2 Structural Sequence
Description: In materials science and molecular biology, thermostability is the ability of a substance to resist irreversible change in its chemical or physical structure, often by resisting decomposition or polymerization, at a high relative temperature.
Number of labels: 1
Problem Type: regression
Columns:
aa_seq: protein amino acid sequence
foldseek_seq: foldseek 20 3di structural sequence
ss8_seq: DSSP 8 secondary… See the full description on the dataset page: https://huggingface.co/datasets/AI4Protein/Thermostability_AlphaFold2.DeepLocBinary_AlphaFold2
DeepLocBinary Dataset with AlphaFold2 Structural Sequence
Description: Protein localization encompasses the processes that establish and maintain proteins at specific locations.
Number of labels: 2
Problem Type: single_label_classification
Columns:
aa_seq: protein amino acid sequence
foldseek_seq: foldseek 20 3di structural sequence
ss8_seq: DSSP 8 secondary structure sequence
location: On the membrane or not
Github
Simple, Efficient and Scalable Structure-aware… See the full description on the dataset page: https://huggingface.co/datasets/AI4Protein/DeepLocBinary_AlphaFold2.GO_MF_AlphaFold2
GO-MF Dataset with AlphaFold2 Structural Sequence
Description: Molecular Function of Gene Ontology (GO) project.
Number of labels: 489
Problem Type: multi_label_classification
Columns:
aa_seq: protein amino acid sequence
foldseek_seq: foldseek 20 3di structural sequence
ss8_seq: DSSP 8 secondary structure sequence
Github
Simple, Efficient and Scalable Structure-aware Adapter Boosts Protein Language Models
https://github.com/tyang816/SES-Adapter
VenusFactory: A Unified… See the full description on the dataset page: https://huggingface.co/datasets/AI4Protein/GO_MF_AlphaFold2.DeepLocMulti_AlphaFold2
DeepLocMulti Dataset with AlphaFold2 Structural Sequence
Description: Protein localization encompasses the processes that establish and maintain proteins at specific locations.
Number of labels: 10
Problem Type: single_label_classification
Columns:
aa_seq: protein amino acid sequence
foldseek_seq: foldseek 20 3di structural sequence
ss8_seq: DSSP 8 secondary structure sequence
location: Specific location
Github
Simple, Efficient and Scalable Structure-aware… See the full description on the dataset page: https://huggingface.co/datasets/AI4Protein/DeepLocMulti_AlphaFold2.GO_BP_AlphaFold2
GO-BP Dataset with AlphaFold2 Structural Sequence
Description: Biological Process of Gene Ontology (GO) project.
Number of labels: 1943
Problem Type: multi_label_classification
Columns:
aa_seq: protein amino acid sequence
foldseek_seq: foldseek 20 3di structural sequence
ss8_seq: DSSP 8 secondary structure sequence
Github
Simple, Efficient and Scalable Structure-aware Adapter Boosts Protein Language Models
https://github.com/tyang816/SES-Adapter
VenusFactory: A… See the full description on the dataset page: https://huggingface.co/datasets/AI4Protein/GO_BP_AlphaFold2.GO_CC_AlphaFold2
GO-CC Dataset with AlphaFold2 Structural Sequence
Description: Cellular Component of Gene Ontology (GO) project.
Number of labels: 320
Problem Type: multi_label_classification
Columns:
aa_seq: protein amino acid sequence
foldseek_seq: foldseek 20 3di structural sequence
ss8_seq: DSSP 8 secondary structure sequence
Github
Simple, Efficient and Scalable Structure-aware Adapter Boosts Protein Language Models
https://github.com/tyang816/SES-Adapter
VenusFactory: A Unified… See the full description on the dataset page: https://huggingface.co/datasets/AI4Protein/GO_CC_AlphaFold2.eSOL_AlphaFold2
eSOL Dataset
Description: Solubility is a fundamental protein property that has important connotations for therapeutics and use in diagnosis.
Number of labels: 1
Problem Type: regression
Columns:
aa_seq: protein amino acid sequence
gene: gene id
Github
VenusFactory: A Unified Platform for Protein Engineering Data Retrieval and Language Model Fine-Tuning
https://github.com/ai4protein/VenusFactory
Citation
Please cite our work if you use our dataset.… See the full description on the dataset page: https://huggingface.co/datasets/AI4Protein/eSOL_AlphaFold2.alphafold-folding-trajectory-functional-stability-coherence-v0.1What this dataset tests
Whether predicted folding trajectories
remain predictive of functional stability
under stress conditions.
Structure alone is not enough.
The path to structure must stay coherent
with real-world stability.
When that relationship breaks
therapeutic proteins fail
in storage
manufacturing
or use.
Required outputs
trajectory_coherence_score
stability_divergence_flag
degradation_horizon_hours
critical_structure_region
stabilization_strategy
Use case
Antibody engineering… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/alphafold-folding-trajectory-functional-stability-coherence-v0.1.SortingSignal_AlphaFold2
SortingSignal Dataset
Description: Protein sorting signal prediction.
Number of labels: 9
Problem Type: multi_label_classification
Columns:
aa_seq: protein amino acid sequence
Github
VenusFactory: A Unified Platform for Protein Engineering Data Retrieval and Language Model Fine-Tuning
https://github.com/ai4protein/VenusFactory
Citation
Please cite our work if you use our dataset.
@article{tan2025venusfactory,
title={VenusFactory: A Unified Platform for… See the full description on the dataset page: https://huggingface.co/datasets/AI4Protein/SortingSignal_AlphaFold2.DeepLoc2Multi_AlphaFold2
DeepLoc2Multi_AlphaFold2 Dataset
Description: Protein localization encompasses the processes that establish and maintain proteins at specific locations.
Number of labels: 2
Problem Type: single_label_classification
Columns:
aa_seq: protein amino acid sequence
detail: meta information
Github
VenusFactory: A Unified Platform for Protein Engineering Data Retrieval and Language Model Fine-Tuning
https://github.com/ai4protein/VenusFactory
Citation
Please cite… See the full description on the dataset page: https://huggingface.co/datasets/AI4Protein/DeepLoc2Multi_AlphaFold2.alphafold_misinterpretation_classifier_v01AlphaFold Misinterpretation Classifier (AMC) v0.1
Purpose
Help models spot when AlphaFold outputs are used to make claims that go beyond their scope.
Teach restraint.
Promote correct boundaries around structural interpretation.
Columns
claim
misinterpretation_type
reason_hint
action
misinterpretation_type examples
function_from_structure: assuming activity from fold
binding_assertion: assuming ligand interaction
metric_confusion: misreading confidence or aligned error
state_fixation:… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/alphafold_misinterpretation_classifier_v01.alphafold_misinterpretation_classifier_v0.2AlphaFold Misinterpretation Classifier
PurposeDetect common ways people overclaim from AlphaFold outputs.
Input fields
protein_context
alphafold_signals
proposed_inference
Required outputReturn one JSON object
misinterpretationyes or no
error_typemust match allowed list
correctionone sentence
Allowed error_type values
no_error
low_confidence_region_overtrust
interdomain_orientation_overclaim
disorder_as_structure
loop_position_overtrust
complex_negation_from_monomer… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/alphafold_misinterpretation_classifier_v0.2.alphafold-interface-coherence-baseline-mapping-v0.1
What this dataset tests
Whether a model can constructa stable baseline mapof protein–protein interface coherence.
This is the intact coupling state.No stress applied.No mutation applied.
The target is the interface basin.
Inputs
interface_residue_countcontact_densitybaseline_deltaG_bindbaseline_kd_nMcontact_stability_scorehotspot_integrity_scoreallosteric_cross_interface_score
Required outputs
interface_coherence_scoresignal_pathsbaseline_failure_margin… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/alphafold-interface-coherence-baseline-mapping-v0.1.alphafold-structure-function-environment-decoupling-v0.1Purpose
Detect when predicted structure
no longer predicts real function
under environmental stress.
Core problem
Proteins can appear stable
in structure predictions
yet lose activity
in real conditions.
This dataset finds
the early decoupling phase.
Inputs
structure confidence
active-site shift
temperature
pH
solvent exposure
binding trend
aggregation signal
Outputs
coherence score
decoupling flag
collapse horizon
critical region
intervention
Evaluation
Binary accuracy
for decoupling… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/alphafold-structure-function-environment-decoupling-v0.1.DeepET_Topt_AlphaFold2
DeepET_Topt Dataset
Description: protein optimum temperature.
Number of labels: 1
Problem Type: regression
Columns:
aa_seq: protein amino acid sequence
ss8_seq: DSSP 8 secondary structure sequence
foldseek_seq: foldseek 20 3di structural sequence
ss3_seq: DSSP 3 secondary structure sequence
Github
VenusFactory: A Unified Platform for Protein Engineering Data Retrieval and Language Model Fine-Tuning
https://github.com/ai4protein/VenusFactory
Citation
Please… See the full description on the dataset page: https://huggingface.co/datasets/AI4Protein/DeepET_Topt_AlphaFold2.alphafold-allosteric-signal-transmission-coherence-mapping-v0.1
What this dataset tests
Whether an intelligence system can detect whendynamic motion in one protein regioncoherently transmits to the active siteor when that communication collapses.
The dataset treats proteins as communication networks.Signal from region A must reach the active sitewith measurable gain and stability.
Loss of transmission predictsallosteric drug failureand resistance emergence.
Required outputs
transmission_coherence_score
active_site_response_gain… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/alphafold-allosteric-signal-transmission-coherence-mapping-v0.1.alphafold-interface-decoherence-under-stress-detection-v0.1
What this dataset tests
Whether a model can detectstress-induced interface decoherencebefore full dissociation.
Stress modes covered
pH shiftheat stressoxidative stress
The failure mode
Contact maps stop predicting bindingand cross-interface signal transmission weakens.
Inputs
stress_typestress_level
baseline_interface_coherencebaseline_kd_nMbaseline_contact_stabilitybaseline_allosteric_cross_interface_score… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/alphafold-interface-decoherence-under-stress-detection-v0.1.alphafold-allosteric-network-fracture-detection-v0.1
Goal
Detect when a mutationfractures the allosteric communication networklinking distal regions to the active site.
Proteins function throughlong-range signal transmission.A mutation can break that signalwithout altering the binding site itself.
This dataset trains systems to detectnetwork fracture beforefunctional collapse appears.
Required outputs
fracture_flag
fracture_horizon_steps
critical_path_loss
network_resilience_index
allosteric_gain_change… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/alphafold-allosteric-network-fracture-detection-v0.1.alphafold-conformation-interaction-coherence-decay-v0.1Goal
Detect when predicted conformation
no longer supports real binding interaction.
Why it matters
Proteins often keep structure
but lose interaction reliability.
Drug discovery fails here.
Design pipelines miss this drift.
Inputs
conformation state
interface shift
environment conditions
binding trend
Outputs
coherence score
decoupling flag
collapse horizon
critical interface
intervention
Evaluation
Flag accuracy
for decoupling detection
MAE
for coherence score
Combined final score
in… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/alphafold-conformation-interaction-coherence-decay-v0.1.VenusVaccine_BacteriaBinary_AlphaFold2VenusVaccine_TumorBinary_AlphaFold2VenusVaccine_VirusBinary_AlphaFold2
