Skip to main content

Table 4 Significance of GO terms from different categories on the performance of mLASSO and mEN based on leave-one-out cross-validation (LOOCV) on the human dataset

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

Measures

mLASSO

mEN

 

All

CC + MF

CC + BP

MF + BP

All

CC + MF

CC + BP

MF + BP

OAA

0.729

0.662

0.654

0.385

0.743

0.621

0.640

0.440

OLA

0.820

0.726

0.715

0.436

0.836

0.686

0.701

0.492

Accuracy

0.814

0.733

0.724

0.446

0.827

0.690

0.709

0.506

Precision

0.859

0.782

0.773

0.500

0.869

0.739

0.760

0.560

Recall

0.857

0.759

0.747

0.457

0.870

0.712

0.730

0.521

F1

0.843

0.758

0.748

0.469

0.855

0.713

0.733

0.528

Micro F1

0.826

0.750

0.741

0.462

0.837

0.711

0.728

0.516

Macro F1

0.638

0.435

0.426

0.212

0.741

0.410

0.427

0.346

HL

0.029

0.041

0.042

0.086

0.028

0.047

0.044

0.078

  1. Note that OAA is the most stringent and objective among all the metrics. CC: cellular components; MF: molecular functions; BP: biological processes. Data in bold represent the best result of the corresponding measures among all predictors