Antimicrobial peptide identification using multi-scale convolutional network

BMC Bioinformatics

Table 8 Comparison of auROC using DeLong’s test on APD3 dataset

Method 1	Method 2	auROC 1	auROC 2	Difference	P value
proposed model	CAMP-DA	0.9892	0.9069	0.0823	< 0.0001
proposed model	CAMP-RF	0.9892	0.9528	0.0365	< 0.0001
proposed model	CAMP-SVM	0.9892	0.9202	0.0690	< 0.0001
proposed model	gkmSVM	0.9892	–	–	NA
proposed model	iAMP-2 L	0.9892	0.8722	0.1170	< 0.0001
proposed model	iAMPpred	0.9892	0.9466	0.0427	< 0.0001
proposed model	DNN	0.9892	0.9802	0.0091	< 0.0001
proposed fusion model	CAMP-DA	0.9918	0.9069	0.0849	< 0.0001
proposed fusion model	CAMP-RF	0.9918	0.9528	0.0391	< 0.0001
proposed fusion model	CAMP-SVM	0.9918	0.9202	0.0716	< 0.0001
proposed fusion model	gkmSVM	0.9918	–	–	NA
proposed fusion model	iAMP-2 L	0.9918	0.8722	0.1196	< 0.0001
proposed fusion model	iAMPpred	0.9918	0.9466	0.0453	< 0.0001
proposed fusion model	DNN	0.9918	0.9802	0.0117	< 0.0001
proposed fusion model	proposed model	0.9918	0.9892	0.0026	< 0.0001

Note: the mark’—’ means that the result is not available. In this experiment, ‘gkmSVM’ method couldn’t be run successfully because the kernel requirement isn’t satisfied

ISSN: 1471-2105