Skip to main content

Table 3 The performace comparision on PDBbind-v2016 core set

From: CurvAGN: Curvature-based Adaptive Graph Neural Networks for Predicting Protein-Ligand Binding Affinity

Methods

RMSE\(\downarrow\)

MAE\(\downarrow\)

SD\(\downarrow\)

R\(\uparrow\)

Free-spatial

GCN

1.735 (0.034)

1.343 (0.037)

1.719 (0.027)

0.613 (0.016)

 

GAT

1.765 (0.026)

1.354 (0.033)

1.740 (0.027)

0.601 (0.016)

Methods

GIN

1.640 (0.044)

1.261 (0.044)

1.621 (0.036)

0.667 (0.018)

 

GAT-GCN

1.562 (0.022)

1.191 (0.016)

1.558(0.018)

0.697 (0.008)

Coordinate

SGCN

1.583 (0.033)

1.250 (0.036)

1.582 (0.320)

0.686 (0.015)

 

MAT

1.457 (0.037)

1.154 (0.037)

1.445 (0.033)

0.747 (0.013)

Distance

GNN-DTI

1.492 (0.025)

1.192 (0.032)

1.471 (0.051)

0.736 (0.021)

Methods

CMPNN

1.408 (0.028)

1.117 (0.031)

1.399 (0.025)

0.765 (0.009)

 

ELGN

1.285 (0.027)

1.013 (0.022)

1.263(0.026)

0.810 (0.012)

Angle

DimeNet

1.453 (0.027)

1.138 (0.026)

1.434 (0.023)

0.752 (0.010)

Methods

SIGN

1.316 (0.031)

1.027 (0.025)

1.312(0.035)

0.797 (0.012)

Ours

CurvAGN

1.217 (0.012)

0.930(0.014)

1.191(0.015)

0.8305 (0.004)

  1. Source: We present the average (standard deviation) across 5 random runs, highlighting the best results. Note that the upward arrow \(\uparrow\) indicates that a higher value is better, while the downward arrow \(\downarrow\) indicates that a higher value is worse