Viral nervous necrosis resistance in gilthead sea bream (Sparus aurata) at the larval stage: heritability and accuracy of genomic prediction with different training and testing settings

Table 4 Average metrics (SD) for classification of VNN symptomatology based on genomic predictions of phenotype or of breeding values as classifiers

Classifier	Method	Metric
Classifier	Method	AUC	ACC	MCC
Genomic-predicted phenotype	Bayes B	0.581 (0.017)	0.569 (0.016)	0.175 (0.024)
	Bayes C	0.597 (0.013)	0.581 (0.013)	0.176 (0.023)
	Bayes Ridge Regression	0.601 (0.013)	0.583 (0.009)	0.178 (0.022)
Genomic-predicted EBV	Bayes B	0.657 (0.002)	0.617 (0.006)	0.237 (0.008)
	Bayes C	0.657 (0.002)	0.617 (0.006)	0.237 (0.008)
	Bayes Ridge Regression	0.657 (0.002)	0.616 (0.005)	0.236 (0.008)

EBV breeding values estimated using the phenotypic information of the animal and of its full- and half-sibs
AUC area under the ROC curve, ACC accuracy as (true positives + true negatives)/number of samples, MCC Matthews correlation coefficient

ISSN: 1297-9686