Skip to main content

Table 1 Performance comparisons on independent test dataset for manifold ranking and SVM-based ranking

From: In-silico prediction of blood-secretory human proteins using a ranking algorithm

AUC

No. of queries

Methods

1

2

3

4

5

Ave.

10

MR

0.7412

0.6565

0.6342

0.6355

0.6576

0.6650

 

SVM-1

0.6342

0.6224

0.6571

0.6317

0.5964

0.6284

 

SVM-2

0.6425

0.6342

0.6521

0.6218

0.6091

0.6319

20

MR

0.7920

0.7629

0.7657

0.7574

0.8046

0.7765

 

SVM-1

0.6768

0.6535

0.6373

0.6371

0.6895

0.6589

 

SVM-2

0.6928

0.6634

0.6823

0.6576

0.6797

0.6752

30

MR

0.8245

0.8283

0.8170

0.8072

0.8655

0.8285

 

SVM-1

0.7388

0.7800

0.8014

0.7864

0.7759

0.7765

 

SVM-2

0.7818

0.8167

0.8023

0.7689

0.7909

0.7921

  1. (Both rankings were performed with 10, 20 and 30 queries, evaluated by AUC of recall-precision curve. (Parameter setting: σ = 2.7003 and α = 0.5. Each kind of query is performed 5 times. MR: Manifold ranking; SVM-1: first strategy of SVM-based ranking; SVM-2: one-class SVM-based ranking)
Back to article page

BMC Bioinformatics

ISSN: 1471-2105

Contact us