Skip to main content

Table 2 Classification accuracy of different classifiers with 560 features for 4 kinase datasets

From: Using multitask classification methods to investigate the kinase-specific phosphorylation sites

 

CDK kinase family

CK2 kinase family

PKA kinase family

PKC kinase family

window size

SVM-rbf

SVM-linear

RF

SVM -rbf

SVM-linear

RF

SVM-rbf

SVM-linear

RF

SVM-rbf

SVM-linear

RF

3

0.8598

0.8613*

0.83

0.7783

0.7796

0.7326

0.6656

0.6678

0.6289

0.6723

0.6758

0.6113

5

0.8013

0.8122

0.7579

0.806

0.8112

0.7935

0.7156

0.7178

0.7111

0.7173

0.724

0.7069

7

0.7578

0.7581

0.7455

0.8655

0.8724

0.8599

0.7567

0.7533*

0.7589

0.7242

0.7196

0.7253

9

0.7305

0.7077

0.7171

0.8706

0.8688

0.8548

0.7456

0.7489

0.7622

0.7253

0.7393*

0.7183

11

0.7223

0.724

0.7226

0.8654

0.8617

0.8619

0.7433

0.7478

0.7511

0.7161

0.7298

0.7023

13

0.721

0.7103

0.7049

0.867

0.8705

0.874

0.7367

0.7378

0.7311

0.7287

0.7299

0.7299

15

0.7211

0.7023

0.7049

0.8724

0.8723

0.8792

0.7322

0.7267

0.7156

0.7264

0.7286

0.7253

17

0.7087

0.7038

0.717

0.8812

0.8811*

0.874

0.7211

0.7167

0.7089

0.7253

0.7286

0.7252

19

0.7142

0.6995

0.7049

0.8759

0.8844

0.8757

0.72

0.6978

0.7167

0.7173

0.7194

0.7194

21

0.7263

0.6887

0.72

0.8759

0.8775

0.8739

0.7133

0.7022

0.7044

0.7082

0.7309

0.6999

  1. *Best performance for each kinase family by SVM with linear kernel and the corresponding window size is selected as the final optimized window size for that kinase family.