Protein Property Prediction on 21 protein landscapes 128 training points (extrapolation) benchmark leaderboard · SOTA2 Research

Protein Property Prediction on 21 protein landscapes 128 training points (extrapolation)

0.669Spearman CorrelationLOCK-GP

Evaluation Results

Method	Links
LOCK-GP2026.06		0.669	0.711	0.592
MLP-SaProt-LastLayer2026.06		0.644	0.64	0.789
Kermut-GP2026.06		0.639	0.654	0.845
Tanimoto-GP2026.06		0.632	0.654	0.762
ConFit2026.06		0.632	0.622	0.882
Ridge-ESM2-8M2026.06		0.629	0.653	0.781
MLP-ESM2-LastLayer2026.06		0.628	0.619	0.771
KermutStruc-GP2026.06		0.624	0.641	0.847
Ridge-ESM2-650M2026.06		0.623	0.649	0.864
MLP-ESM2-8M2026.06		0.621	0.648	0.711
Ridge-OH-ESM2-650M-Aug2026.06		0.62	0.624	0.909
MLP-OH-ESM2-650M-Aug2026.06		0.619	0.632	0.789
Ridge-OH2026.06		0.589	0.599	0.939
SigGLM-OH2026.06		0.587	0.61	0.8
MLP-ESM2-8M-RandInit2026.06		0.581	0.609	0.747
MLP-OH2026.06		0.577	0.596	0.863
MLP-ESM2-650M2026.06		0.568	0.576	0.767
Ridge-ESM2-8M-MeanPool2026.06		0.552	0.564	0.886
MLP-ESM2-8M-MeanPool2026.06		0.536	0.553	0.814
KermutSeq-GP2026.06		0.518	0.505	0.836
BLOSUM50-ZeroShot2026.06		0.27	0.256	—
ESM2-650M-ZeroShot2026.06		-0.072	-0.063	—