Predict Module
Create predictions from protein sequences using pretrained models.
The predict module provides sequence-level, residue-level, and mutation-level inference for protein analysis.
Model catalog overview
| Model | Primary task | Protocol | Output granularity |
|---|---|---|---|
| TMbed | Transmembrane topology prediction | residue_to_class | Per residue |
| LightAttentionMembrane | Membrane association classification | residues_to_class | Per sequence |
| LightAttentionSubcellularLocalization | Subcellular localization classification | residues_to_class | Per sequence |
| Seth | Disorder scoring (CheZOD-based) | residue_to_value | Per residue |
| UdonPred | Disorder scoring (TriZOD) | residue_to_value | Per residue |
| BindEmbed | Ligand-binding residue prediction | residue_to_class | Per residue |
| ProtT5Conservation | Residue conservation estimation | residue_to_class | Per residue |
| ProtT5SecondaryStructure | Secondary structure prediction (3-state and 8-state) | residue_to_class | Per residue |
| VespaG | Variant effect prediction | residue_to_class | Mutation |
| ExoTox | Exotoxin classification | sequence_to_class | Per sequence |