Table 2  Evaluation metrics for AiKPro model on strict split subsets.

From: AiKPro: deep learning model for kinome-wide bioactivity profiling using structure-based sequence alignments and molecular 3D conformer ensemble descriptors

Test data

Count

Variance

PCC

R2

RMSE

MAE

Strict split for kinases

1522

1.50

0.71

0.47

0.89

0.72

CLK1

450

1.02

0.69

0.29

0.85

0.73

PRKCE

488

1.71

0.71

0.48

0.94

0.78

HCK

584

1.59

0.77

0.55

0.85

0.65

Strict split for compounds

15,271

0.56

0.87

0.73

0.39

0.29

Strict split for docking

563

0.63

0.89

0.78

0.38

0.28

Back to article page