Skip to main content

Table 3 RMSE performance of each model on the four tasks

From: A multitask GNN-based interpretable model for discovery of selective JAK inhibitors

 

Methods

Training set

Validation set

Test set

Global

MTATFP

0.23

0.51

0.52

STATFP

0.29

0.56

0.57

LightGBM_MD

0.30

0.62

0.61

LightGBM_ECFP4

0.33

0.60

0.60

JAK1

MTATFP

0.27

0.52

0.53

STATFP

0.27

0.51

0.55

LightGBM_MD

0.34

0.61

0.60

LightGBM_ECFP4

0.35

0.57

0.60

JAK2

MTATFP

0.26

0.51

0.54

STATFP

0.28

0.55

0.56

LightGBM_MD

0.31

0.66

0.62

LightGBM_ECFP4

0.36

0.63

0.64

JAK3

MTATFP

0.21

0.58

0.58

STATFP

0.36

0.70

0.66

LightGBM_MD

0.33

0.67

0.65

LightGBM_ECFP4

0.38

0.66

0.62

TYK2

MTATFP

0.20

0.44

0.42

STATFP

0.26

0.49

0.52

LightGBM_MD

0.21

0.52

0.55

LightGBM_ECFP4

0.25

0.52

0.52

  1. The lower the RMSE value was, the better the model performed