Protein Property Prediction on 21 protein property datasets 128 data points (Extrapolation)
6.81Spearman CorrelationKermutSeq-GP
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| KermutSeq-GPtype=Sequence2026.06 | 6.81 | 7.29 | 6.19 | |
| SigGLM-OHRepresentation=One-Hot2026.06 | 6.14 | 5.9 | 5.52 | |
| Ridge-OHRepresentation=One-Hot2026.06 | 5.95 | 6.14 | 6 | |
| KermutStruc-GPtype=Structure2026.06 | 5.19 | 4.86 | 5.19 | |
| Ridge-ESM2Backbone=ESM22026.06 | 5.19 | 4.67 | 4.57 | |
| Tanimoto-GP2026.06 | 4.48 | 3.86 | 5.48 | |
| Kermut-GP2026.06 | 4.33 | 4.29 | 4.24 | |
| MLP-ESM2-LastLayerBackbone=ESM2, Layer=LastLayer2026.06 | 4.05 | 5.62 | 5.19 | |
| LOCK-GP2026.06 | 2.86 | 2.38 | 2.62 | |
| LOCK-GPModel Class=Gaussian process, Prior Information=BLOSUM, Number of training data points=128, Evaluation regime=Extrapolation2026.06 | 0.669 | 0.711 | 0.592 | |
| Kermut-GPModel Class=Gaussian process, Prior Information=ESM-2 + ProteinMPNN, Number of training data points=128, Evaluation regime=Extrapolation2026.06 | 0.639 | 0.654 | 0.845 | |
| Tanimoto-GPModel Class=Gaussian process, Prior Information=BLOSUM, Number of training data points=128, Evaluation regime=Extrapolation2026.06 | 0.632 | 0.654 | 0.762 | |
| Ridge-ESM2Model Class=Neural network, Prior Information=ESM-2, Number of training data points=128, Evaluation regime=Extrapolation2026.06 | 0.629 | 0.653 | 0.781 | |
| MLP-ESM2-LastLayerModel Class=Neural network, Prior Information=ESM-2, Number of training data points=128, Evaluation regime=Extrapolation2026.06 | 0.628 | 0.619 | 0.771 | |
| KermutStruc-GPModel Class=Gaussian process, Prior Information=ESM-2 + ProteinMPNN, Number of training data points=128, Evaluation regime=Extrapolation2026.06 | 0.624 | 0.641 | 0.847 | |
| Ridge-OHModel Class=Linear, Prior Information=—, Number of training data points=128, Evaluation regime=Extrapolation2026.06 | 0.589 | 0.599 | 0.939 | |
| SigGLM-OHModel Class=Linear + global non-linearity, Prior Information=—, Number of training data points=128, Evaluation regime=Extrapolation2026.06 | 0.587 | 0.61 | 0.8 | |
| KermutSeq-GPModel Class=Gaussian process, Prior Information=ESM-2, Number of training data points=128, Evaluation regime=Extrapolation2026.06 | 0.518 | 0.505 | 0.836 |