Protein Property Prediction on 21 protein property datasets 512 data points (Extrapolation)
7.24Spearman CorrelationKermutSeq-GP
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| KermutSeq-GPtype=Sequence2026.06 | 7.24 | 6.81 | 6.43 | |
| Ridge-OHRepresentation=One-Hot2026.06 | 6.33 | 7 | 7 | |
| SigGLM-OHRepresentation=One-Hot2026.06 | 6.14 | 5.86 | 5.86 | |
| KermutStruc-GPtype=Structure2026.06 | 5.43 | 6.19 | 5.76 | |
| Ridge-ESM2Backbone=ESM22026.06 | 5.1 | 5 | 5.33 | |
| Tanimoto-GP2026.06 | 4.38 | 3.71 | 3.43 | |
| Kermut-GP2026.06 | 4.29 | 3.57 | 4.43 | |
| MLP-ESM2-LastLayerBackbone=ESM2, Layer=LastLayer2026.06 | 3.48 | 4.43 | 4.76 | |
| LOCK-GP2026.06 | 2.62 | 2.43 | 2 | |
| LOCK-GPModel Class=Gaussian process, Prior Information=BLOSUM, Number of training data points=512, Evaluation regime=Extrapolation2026.06 | 0.759 | 0.807 | 0.439 | |
| Kermut-GPModel Class=Gaussian process, Prior Information=ESM-2 + ProteinMPNN, Number of training data points=512, Evaluation regime=Extrapolation2026.06 | 0.75 | 0.767 | 0.67 | |
| Tanimoto-GPModel Class=Gaussian process, Prior Information=BLOSUM, Number of training data points=512, Evaluation regime=Extrapolation2026.06 | 0.739 | 0.769 | 0.574 | |
| MLP-ESM2-LastLayerModel Class=Neural network, Prior Information=ESM-2, Number of training data points=512, Evaluation regime=Extrapolation2026.06 | 0.738 | 0.742 | 0.618 | |
| Ridge-ESM2Model Class=Neural network, Prior Information=ESM-2, Number of training data points=512, Evaluation regime=Extrapolation2026.06 | 0.719 | 0.74 | 0.685 | |
| KermutStruc-GPModel Class=Gaussian process, Prior Information=ESM-2 + ProteinMPNN, Number of training data points=512, Evaluation regime=Extrapolation2026.06 | 0.706 | 0.713 | 0.847 | |
| SigGLM-OHModel Class=Linear + global non-linearity, Prior Information=—, Number of training data points=512, Evaluation regime=Extrapolation2026.06 | 0.697 | 0.724 | 0.669 | |
| Ridge-OHModel Class=Linear, Prior Information=—, Number of training data points=512, Evaluation regime=Extrapolation2026.06 | 0.694 | 0.698 | 0.877 | |
| KermutSeq-GPModel Class=Gaussian process, Prior Information=ESM-2, Number of training data points=512, Evaluation regime=Extrapolation2026.06 | 0.605 | 0.614 | 0.714 |