Skip to main content

Table 1 Comparing mLASSO and mEN with state-of-the-art multi-label predictors based on leave-one-out cross-validation on the human dataset

From: Sparse regressions for predicting and interpreting subcellular localization of multi-label proteins

Label

Subcellular location

LOOCV Locative Accuracy (LA)

  

iLoc-Hum [68]

mGOASVM [44]

mLASSO

mEN

1

Centrosome

56/77 = 0.727

64/77 = 0.831

42/77 = 0.546

60/77 = 0.779

2

Cytoplasm

561/817 = 0.687

683/817 = 0.836

699/817 = 0.856

683/817 = 0.836

3

Cytoskeleton

27/79 = 0.342

44/79 = 0.557

29/79 = 0.367

32/79 = 0.405

4

Endoplasmic reticulum

166/229 = 0.725

193/229 = 0.843

194/229 = 0.847

190/229 = 0.830

5

Endosome

1/24 = 0.042

9/24 = 0.375

1/24 = 0.042

5/24 = 0.208

6

Extracellular

325/385 = 0.844

344/385 = 0.894

311/385 = 0.808

314/385 = 0.816

7

Golgi apparatus

99/161 = 0.615

131/161 = 0.814

118/161 = 0.733

128/161 = 0.795

8

Lysosome

56/77 = 0.727

71/77 = 0.922

62/77 = 0.805

74/77 = 0.961

9

Microsome

7/24 = 0.292

18/24 = 0.750

1/24 = 0.042

14/24 = 0.583

10

Mitochondrion

284/364 = 0.780

339/364 = 0.931

336/364 = 0.923

336/364 = 0.923

11

Nucleus

918/1021 = 0.899

931/1021 = 0.912

922/1021 = 0.903

923/1021 = 0.904

12

Peroxisome

20/47 = 0.426

43/47 = 0.915

34/47 = 0.723

39/47 = 0.830

13

Plasma membrane

277/354 = 0.783

288/354 = 0.814

267/354 = 0.754

266/354 = 0.751

14

Synapse

12/22 = 0.546

12/22 = 0.546

3/22 = 0.136

13/22 = 0.591

Overall Actual Accuracy (OAA)

2118/3106 = 0.682

2251/3106 = 0.725

2265/3106 = 0.729

2307/3106 = 0.743

Overall Locative Accuracy (OLA)

2809/3681 = 0.763

3170/3681 = 0.861

3019/3681 = 0.820

3077/3681 = 0.836

Accuracy

0.821

0.814

0.827

Precision

0.851

0.859

0.869

Recall

0.888

0.857

0.870

F1

0.853

0.843

0.855

Micro F1

0.835

0.826

0.837

Macro F1

0.740

0.638

0.741

HL

0.029

0.029

0.028

  1. “–” means the corresponding references do not provide the related metrics. Note that OAA is the most stringent and objective among all the metrics. Data in bold represent the best result of the corresponding measures among all predictors