Characterization of the prohormone complement in cattle using genomic libraries and cleavage prediction approaches

Table 3 Accuracy of different cleavage prediction models developed on cattle neuropeptides

	Cleavage Prediction Model
	AA¹		AA-Prop²
Model Accuracy Statistic	LR³	ANT⁴	LR	ANT
True Negative	669	691	667	690
True Positive	44	43	50	46
False Negative	41	42	35	39
False Positive	77	55	79	56
Correct Classification Rate (%)	86	88	86	89
Sensitivity (%)	52	51	59	54
Specificity (%)	90	93	89	92
Area under receiver operating characteristic curve (%)	76	76	82	78

¹AA = Models developed using only amino acids
² AA-Prop = Models developed using amino acids plus amino acid physiochemical properties
³LR = Models developed using logistic regression
⁴ ANT = Models developed using artificial neural networks

ISSN: 1471-2164