Skip to main content

Table 2 The experimental results of various individual and combinative features on the training set for general BLPs

From: Prediction of bioluminescent proteins by using sequence-derived features and lineage-specific scheme

Type

Feature

Sensitivity

Specificity

Accuracy

MCC

AUC

Individual

AACa

0.729 ± 0.029

0.806 ± 0.023

0.767 ± 0.020

0.537 ± 0.039

0.802 ± 0.012

DCb

0.791 ± 0.017

0.857 ± 0.028

0.824 ± 0.014

0.650 ± 0.029

0.830 ± 0.016

MTFc

0.313 ± 0.017

0.942 ± 0.012

0.628 ± 0.008

0.328 ± 0.018

0.653 ± 0.010

PCPd

0.452 ± 0.010

0.910 ± 0.026

0.681 ± 0.010

0.408 ± 0.029

0.763 ± 0.014

Combinative

AAC + DC

0.799 ± 0.015

0.862 ± 0.026

0.830 ± 0.008

0.663 ± 0.018

0.841 ± 0.012

AAC + MTF

0.764 ± 0.016

0.801 ± 0.021

0.783 ± 0.005

0.566 ± 0.011

0.810 ± 0.007

AAC + PCP

0.728 ± 0.013

0.809 ± 0.013

0.768 ± 0.007

0.538 ± 0.015

0.813 ± 0.008

DC + MTF

0.799 ± 0.014

0.854 ± 0.008

0.826 ± 0.008

0.653 ± 0.015

0.836 ± 0.005

DC + PCP

0.775 ± 0.014

0.878 ± 0.020

0.827 ± 0.004

0.658 ± 0.009

0.841 ± 0.006

MTF + PCP

0.477 ± 0.010

0.917 ± 0.016

0.697 ± 0.004

0.440 ± 0.014

0.764 ± 0.020

AAC + DC + MTF

0.772 ± 0.007

0.888 ± 0.011

0.830 ± 0.008

0.665 ± 0.016

0.842 ± 0.006

AAC + DC + PCP

0.780 ± 0.007

0.880 ± 0.016

0.830 ± 0.009

0.663 ± 0.019

0.845 ± 0.009

AAC + MTF + PCP

0.742 ± 0.011

0.793 ± 0.004

0.767 ± 0.005

0.536 ± 0.010

0.816 ± 0.004

DC + MTF + PCP

0.775 ± 0.014

0.886 ± 0.025

0.830 ± 0.011

0.665 ± 0.023

0.845 ± 0.014

AAC + DC + MTF + PCP

0.770 ± 0.010

0.894 ± 0.014

0.836 ± 0.004

0.676 ± 0.010

0.850 ± 0.006

  1. The results are reported by maximizing the MCC value of prediction on the corresponding dataset over five-fold cross-validation. a indicates the features of amino acid composition; b stands for the features of dipeptide composition; c is the features of motifs; d represents the features of physicochemical properties