Skip to main content

Table 4 The performance comparison of models in protein centric task. The table reports the true positive (TP), false negative (FN), false positive (FP), number of proteins that have at least 1 prediction label (NP), the precision, recall, F measure, and Matthews correlation coefficient (MCC) for different features used in the models. The features used are k-mer, GO terms (BP, CC, and MF), and when both k-mer and GO terms are combined. The number of layers in neural network are three where the dimension of the first two layers are 128 and 64, and the last layer dimension is equal to the number of metabolic pathways

From: Predicting metabolic pathway membership with deep neural networks by integrating sequential and ontology information

Features

Method

TP

FN

FP

NP

Precision

Recall

F1 measure

MCC

k-mer

NN

1312

4183

3481

920

0.216

0.178

0.195

0.246

 

SVM

206

5289

41

149

0.823

0.074

0.136

0.176

 

RF

201

5294

0

71

1.000

0.052

0.099

0.190

 

KNN

792

4703

457

332

0.624

0.132

0.217

0.300

BP

NN

2773

2722

2256

984

0.646

0.513

0.572

0.521

 

SVM

2283

3212

711

888

0.796

0.449

0.574

0.560

 

RF

1709

3786

170

623

0.906

0.375

0.531

0.529

 

KNN

1391

4104

424

787

0.755

0.374

0.500

0.438

BP+k-mer

NN

2760

2735

1830

973

0.648

0.500

0.565

0.544

 

SVM

2301

3194

709

883

0.804

0.448

0.575

0.563

 

RF

1198

4297

42

371

0.970

0.227

0.368

0.456

 

KNN

1430

4065

394

773

0.764

0.375

0.503

0.449

CC

NN

1945

3550

3616

926

0.422

0.355

0.386

0.343

 

SVM

1117

4378

283

534

0.768

0.255

0.383

0.400

 

RF

1009

4486

267

445

0.736

0.215

0.333

0.379

 

KNN

966

4529

436

525

0.627

0.204

0.308

0.346

CC+k-mer

NN

2178

3317

2523

827

0.493

0.348

0.408

0.421

 

SVM

1213

4282

302

551

0.784

0.270

0.401

0.418

 

RF

659

4836

25

212

0.977

0.132

0.232

0.338

 

KNN

1026

4469

450

535

0.675

0.224

0.336

0.358

MF

NN

2429

3066

2950

844

0.545

0.400

0.462

0.439

 

SVM

1703

3792

423

646

0.785

0.306

0.441

0.496

 

RF

1580

3915

454

604

0.786

0.316

0.451

0.470

 

KNN

1313

4182

576

635

0.642

0.262

0.372

0.405

MF+k-mer

NN

2520

2975

2900

868

0.580

0.399

0.472

0.454

 

SVM

1771

3724

449

665

0.783

0.326

0.460

0.504

 

RF

985

4510

18

275

0.968

0.157

0.270

0.417

 

KNN

1427

4068

533

612

0.697

0.272

0.391

0.432