MultiToxPred 1.0: a novel comprehensive tool for predicting 27 classes of protein toxins using an ensemble machine learning approach

Table 5 Ten-fold cross-validation on the training and testing datasets using the PAAC and DPC molecular descriptors via ensemble algorithms

Ensemble algorithms	ACC	F1	PPV	TPR	AUC
PAAC
RF + LightGBM ^{Training−CV}	0.789	0.781	0.779	0.788	0.98
RF + LightGBM ^Testing	0.800	0.796	0.793	0.803	0.99
DPC
MLP + LightGBM ^{Training−CV}	0.813	0.801	0.801	0.812	0.99
MLP + LightGBM ^Testing	0.816	0.806	0.799	0.820	0.99
MLP + QDA^{Training−CV}	0.831	0.817	0.822	0.830	0.99
MLP + QDA^Testing	0.844	0.837	0.843	0.847	0.99
LightGBM + QDA^{Training−CV}	0.840*	0.827*	0.836*	0.840*	0.99*
LightGBM + QDA^Testing	0.846*	0.838*	0.847*	0.849*	0.99*

ISSN: 1471-2105