Table 2 Performance evaluation on regression datasets from MoleculeNet

From: MolGraph-xLSTM as a graph-based dual-level xLSTM framework for enhanced molecular representation and interpretability

 

ESOL

Lipo

Freesolv

 

RMSE

PCC

RMSE

PCC

RMSE

PCC

FP-GNN

0.658 ± 0.006

0.946 ± 0.006

0.610 ± 0.028

0.861 ± 0.012

1.106 ± 0.195

0.951 ± 0.023

DeeperGCN

0.615 ± 0.044

0.954 ± 0.008

0.645 ± 0.048

0.842 ± 0.026

1.261 ± 0.022

0.938 ± 0.007

DMPNN

0.575 ± 0.073

0.957 ± 0.015

0.553 ± 0.033

0.842 ± 0.026

1.211 ± 0.120

0.945 ± 0.007

HiGNN

0.570 ± 0.061

0.959 ± 0.013

0.563 ± 0.041

0.882 ± 0.018

1.068 ± 0.092

0.956 ± 0.007

TransFoxMol

0.930 ± 0.261

0.917 ± 0.047

0.652 ± 0.033

0.855 ± 0.011

1.225 ± 0.155

0.945 ± 0.007

BiLSTM

0.743 ± 0.038

0.931 ± 0.012

0.779 ± 0.031

0.765 ± 0.026

1.398 ± 0.070

0.923 ± 0.015

AutoML

0.843 ± 0.062

0.910 ± 0.023

0.792 ± 0.043

0.748 ± 0.031

1.235 ± 0.220

0.941 ± 0.024

MolGraph- xLSTM (Ours)

0.527 ± 0.046

0.965 ± 0.010

0.550 ± 0.026

0.888 ± 0.011

1.024 ± 0.076

0.960 ± 0.006

  1. Best results are shown in bold.