Skip to main content

Table 9 Best hybrid results for the LSTM model on the test set. The bold values show the best results

From: Machine learning-based approaches for ubiquitination site prediction in human proteins

Features

Metrics

Window size

PSSM

AAC

DPC

PCP

Accuracy

Macro precision

Macro recall

Macro-F1

F1positive

MCC

–

–

–

–

82.717 ± 0.917

0.57 ± 0.001

0.581 ± 0.011

0.574 ± 0.004

0.9 ± 0.003

0.396

77

â– 

–

–

–

81.984 ± 2.028

0.572 ± 0.005

0.59 ± 0.023

0.576 ± 0.007

0.902 ± 0.005

0.402

45

–

â– 

–

–

82.036 ± 1.282

0.568 ± 0.001

0.583 ± 0.011

0.572 ± 0.003

0.896 ± 0.004

0.398

77

–

–

â– 

–

82.73 ± 0.008

0.566 ± 0.002

0.574 ± 0.005

0.569 ± 0.003

0.897 ± 0.003

0.396

99

–

–

–

â– 

82.006 ± 1.026

0.568 ± 0.003

0.584 ± 0.005

0.574 ± 0.002

0.901 ± 0.002

0.397

99

â– 

â– 

â– 

â– 

83.13 ± 0.658

0.567 ± 0.002

0.572 ± 0.007

0.569 ± 0.003

0.889 ± 0.002

0.395

77

  1. (â– ) When combined with the sequence feature
  2. (–) When not combined with sequence feature