Skip to main content

Table 1 ARI values of the five clustering algorithms under the four preprocessing methods

From: Impact of data preprocessing on cell-type clustering based on single-cell RNA-seq data

 

Darmanis

Lake

Yan

Baron

Biase

Leng

Romanov

Deng

dynamicTreecut

 sctransform

0.368

0.195

0.667

0.376

*

0.069

0.562

0.858

 log

0.37

0.206

0.667

0.629

1

0.101

0.51

0.858

 no

0.28

0.21

0.667

0.16

0.71

0.052

0.191

0.582

 z-score

0.015

0.003

0.296

0

0.241

0

0

0.018

pcaReduce

 sctransform

0.457

0.292

0.818

0.425

*

0.265

0.379

0.361

 log

0.46

0.29

0.779

0.415

0.388

0.258

0.364

0.409

 no

0.071

0.156

0.41

0.278

0.004

0.154

0.166

0.422

 z-score

0.442

0.276

0.671

0.238

0.325

0.056

0.221

0.518

tSNE + k-means

 sctransform

0.467

0.304

0.679

0.462

*

0.059

0.429

0.481

 log

0.479

0.304

0.684

0.445

0.772

0.055

0.44

0.449

 no

0.43

0.276

0.618

0.311

0.76

0.126

0.29

0.435

 z-score

0.034

0.01

0.351

0.03

0.002

0.015

0.098

0.156

SNN-clip

 sctransform

0.572

0.52

0.673

0.515

*

0.26

0.422

0.594

 log

0.609

0.501

0.673

0.477

0.581

0.241

0.387

0.596

 no

0.077

0.249

0.722

0.326

0.476

0.252

0.279

0.576

 z-score

0.643

0.498

0.744

0.395

0.216

0.277

0.228

0.483

SC3

 sctransform

0.795

0.556

0.658

0.537

*

0.21

0.519

0.571

 log

0.785

0.554

0.674

0.56

0.87

0.22

0.511

0.575

 no

0.492

0.415

0.595

0.757

0.783

0.221

0.575

0.841

 z-score

0.656

0.494

0.895

0.489

0.956

0.594

0.336

0.686

  1. Each bold number in the table represents the maximum ARI of a clustering algorithm.