A meta-learning approach for B-cell conformational epitope prediction

BMC Bioinformatics

Table 12 Ablation analysis of influence of linear base learners on stacking meta classifiers

Classifier^*	TPR	FPR	Precision	Accuracy	F-score	MCC	AUC
Conformational 3-level Stacking	0.144	0.018	0.261	0.946	0.186	0.168	0.801
+BCPREDS	0.184	0.006	0.597	0.960	0.281	0.317	0.753
+AAP	0.194	0.008	0.527	0.958	0.284	0.303	0.788
+ BepiPred	0.214	0.009	0.524	0.958	0.304	0.317	0.788
+ABCpred	0.194	0.008	0.520	0.958	0.283	0.300	0.793

^*Classifiers tested in the ablation analysis. The first classifier in the first row is the stacking meta classifier that only employs the 4 conformational base learners. The remaining classifiers are listed in the order in which they were selected to be added iteratively to the stacking meta classifier for the ablation study. `+’ indicates “added.” For example, the second classifier is the stacking meta classifier after BCPREDS was added, and the third classifier is the stacking meta classifier after BCPREDS and AAP were added to the meta model. The meta classifier in the final row applied all the base learners after the final prediction tool ABCpred was added.

ISSN: 1471-2105