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

BMC Bioinformatics

Table 4 Comparison of lineage-specific models with traditional universal models on three training sets using five-fold cross-validation

Lineage	Model	Sensitivity	Specificity	Accuracy	MCC	AUC
Bacteria	PredBLP-U	0.790 ± 0.010	0.918 ± 0.014	0.854 ± 0.003	0.714 ± 0.007	0.872 ± 0.006
Bacteria	PredBLP-B	0.832 ± 0.012	0.943 ± 0.016	0.888 ± 0.006	0.780 ± 0.013	0.920 ± 0.010
Eukaryota	PredBLP-U	0.417 ± 0.053	0.883 ± 0.041	0.650 ± 0.033	0.340 ± 0.075	0.670 ± 0.017
Eukaryota	PredBLP-E	0.667 ± 0.053	0.833 ± 0.053	0.750 ± 0.026	0.510 ± 0.054	0.806 ± 0.015
Archaea	PredBLP-U	0.750 ± 0.079	0.875 ± 0.079	0.813 ± 0.040	0.637 ± 0.081	0.868 ± 0.016
Archaea	PredBLP-A	0.825 ± 0.061	0.900 ± 0.094	0.863 ± 0.047	0.733 ± 0.095	0.917 ± 0.019

The results are reported by maximizing the MCC values of prediction on the corresponding dataset over five-fold cross-validation. PredBLP-U stands for the universal model of the proposed PredBLP predictor. PredBLP-B, PredBLP-E and PredBLP-A indicate three lineage-specific models (i.e. bacteria-, eukaryota- and archaea- specific model) respectively

ISSN: 1471-2105