Skip to main content

Table 2 Different model performances on the Lipo Data Set

From: LogD7.4 prediction enhanced by transferring knowledge from chromatographic retention time, microscopic pKa and logP

Model

R2

Lipo (N = 4200)

MAE

RMSE

CALlogD

− 0.251

0.869

1.356

MolMapNet

0.685 ± 0.036

0.501 ± 0.025

0.682 ± 0.040

MGA

0.768 ± 0.030

0.423 ± 0.022

0.585 ± 0.044

StructGNN

0.791 ± 0.020

0.374 ± 0.019

0.556 ± 0.035

KEMPNN

0.767 ± 0.027

0.410 ± 0.018

0.589 ± 0.040

CoMPT

0.767 ± 0.032

0.417 ± 0.020

0.588 ± 0.046

ALipSol

0.813 ± 0.028

0.362 ± 0.019

0.526 ± 0.048

ALipSol + 

0.820 ± 0.025

0.349 ± 0.021

0.516 ± 0.045

RTlogD

0.835 ± 0.026

0.341 ± 0.018

0.505 ± 0.044

  1. Values in bold represent the superior performance among the various methods
Back to article page

Journal of Cheminformatics

ISSN: 1758-2946

Contact us