Skip to main content

Table 3 Performance comparison. Here, protein complexes are detected from three PPI datasets and MIPS is used as benchmark

From: An effective approach to detecting both small and large complexes from protein-protein interaction networks

Methods

Small

Large

Total

 

Recall

Precision

F1

Recall

Precision

F1

Recall

Precision

F1

(a) Gavin et al.

         

MCODE

0.058

0.143

0.083

0.371

0.411

0.390

0.311

0.392

0.347

RNSC

0.184

0.071

0.102

0.524

0.393

0.449

0.487

0.221

0.304

DPClus

0.165

0.114

0.135

0.529

0.331

0.408

0.495

0.267

0.347

CORE

0.291

0.109

0.159

0.135

0.452

0.208

0.330

0.245

0.281

ClusterONE

0.087

0.276

0.133

0.488

0.481

0.485

0.403

0.492

0.443

CPredictor

0.117

0.212

0.150

0.506

0.421

0.459

0.458

0.437

0.447

CPredictor2.0

0.350

0.146

0.206

0.553

0.627

0.588

0.56

0.418

0.479

(b) Krogan et al.

         

MCODE

0.039

0.120

0.059

0.288

0.511

0.368

0.234

0.431

0.304

RNSC

0.408

0.063

0.110

0.394

0.390

0.392

0.549

0.147

0.233

DPClus

0.369

0.074

0.123

0.418

0.352

0.382

0.549

0.169

0.258

CORE

0.456

0.088

0.147

0.047

0.250

0.079

0.377

0.165

0.229

ClusterONE

0.184

0.088

0.119

0.441

0.132

0.203

0.495

0.176

0.259

CPredictor

0.233

0.132

0.169

0.453

0.425

0.438

0.513

0.338

0.407

CPredictor2.0

0.417

0.158

0.229

0.488

0.610

0.543

0.604

0.391

0.475

(c) Collins et al.

         

MCODE

0.058

0.139

0.082

0.459

0.560

0.504

0.388

0.532

0.449

RNSC

0.398

0.139

0.206

0.471

0.533

0.500

0.590

0.298

0.396

DPClus

0.350

0.146

0.206

0.512

0.440

0.473

0.579

0.313

0.407

CORE

0.388

0.141

0.206

0.247

0.605

0.351

0.451

0.298

0.358

ClusterONE

0.350

0.187

0.244

0.553

0.431

0.484

0.586

0.346

0.435

CPredictor

0.272

0.212

0.238

0.524

0.509

0.516

0.546

0.430

0.481

CPredictor2.0

0.427

0.200

0.272

0.588

0.665

0.624

0.634

0.466

0.537

  1. Each bold value means the largest performance measure among the compared methods on the given PPI dataset