Table 2 Performance comparison on CASF-2007, CASF-2013, and CASF-2016 datasets
From: Harnessing pre-trained models for accurate prediction of protein-ligand binding affinity
Method | CASF-2007 | CASF-2013 | CASF-2016 | ||
|---|---|---|---|---|---|
Pearson \(\uparrow\) | Pearson \(\uparrow\) | RMSE \(\downarrow\) | Pearson \(\uparrow\) | Spearman \(\uparrow\) | |
AutodockVina [42] | – | 0.54 | – | 0.604 | 0.528 |
Glide-SP [7] | 0.343 | 0.452 | – | 0.513 | 0.419 |
Glide-XP [43] | 0.457 | 0.277 | – | 0.467 | 0.257 |
ECIF [44] | – | – | 1.169 | 0.866 | – |
CAPLA [45] | – | 0.770 | 1.200 | 0.843 | – |
SVSBI [46] | – | – | – | 0.832 | – |
KDEEP [9] | – | – | – | 0.701 | 0.528 |
Pafnucy [10] | – | 0.70 | 1.42 | 0.78 | – |
OnionNet-2 [47] | – | 0.821 | 1.164 | 0.864 | – |
GenScore [48] | – | – | – | 0.837 | 0.682 |
ConBAP [49] | – | – | 1.127 | 0.864 | 0.719 |
PIGNet2 [50] | – | – | – | 0.747 | 0.651 |
TopoFormer-Seq [51] | 0.836 | 0.817 | – | 0.865 | – |
SableBind | 0.826 | 0.787 | 1.205 | 0.832 | 0.825 |