Skip to main content

Table 3 Performance comparison of SMPLIP-Score with reported models on the PDBbind v.2015 dataset

From: SMPLIP-Score: predicting ligand binding affinity from simple and interpretable on-the-fly interaction fingerprint pattern descriptors

ML/DL Method Core Set Refs
SMPLIP-Score 0.771 (1.489)
ACNN 0.669 Gomes et al. [23]
TopBP-ML 0.797 (1.99) Cang et al. [57]
TopBP-DL 0.799 (1.91)
RI-Score 0.782 (2.051) Nguyen et al. [58]
Pafnucy 0.70 (1.62) Stepniewska Dziubinska et al. [28]
PLEC-Linear 0.757 (1.47)a Wójcikowski et al. [32]
PLEC-NN 0.774 (1.43)a Wójcikowski et al. [32]
OnionNet 0.782 (1.503) Zheng et al. [31]
RF-Score-v3 0.74 (1.51)a Wójcikowski et al. [59]
X-Score 0.614 (1.78)a Khamis et al. [60]
Autodock Vina 0.54 (1.90)a Gaillard et al. [61]
Autodock 0.54 (1.91)a
  1. Pearson correlation coefficients with RMSE in parentheses for predictions by different methods
  2. SMPLIP-Score interaction fingerprint pattern and Ligand Fragment-based random forest (RF) model, GRID-RF grid featurizer based Random Forest; ACNN Atomic Convolutional Neural Network model, TopBP Topology based model, RI-Score Rigidity Index based score
  3. aValues in parenthesis represent the standard deviation (SD)