Skip to main content

Table 1 Average classification accuracy of different classifiers with 560 features

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

window size

LS-SVMs

MTL-Feat3

MTLS-SVMs

3

0.7381

0.727

0.728

5

0.754

0.7462

0.7459

7

0.7611

0.7595

0.7595

9

0.7498

0.741

0.74

11

0.7504

0.7455

0.7478

13

0.7491

0.7403

0.7416

15

0.7439

0.7355

0.7394

17

0.7439

0.729

0.7316

19

0.7325

0.7251

0.727

21

0.7325

0.7192

0.7176

opt*

0.7939

0.791

0.7936

  1. Five fold cross validation and grid fitting of parameters are used to estimate the performance of all classifiers. *The optimized window sizes (3, 17, 7 and 9) for 4 kinase family datasets are used to build classifiers.
Back to article page

Proteome Science

ISSN: 1477-5956