Strength in numbers: achieving greater accuracy in MHC-I binding prediction by combining the results from multiple prediction tools

Table 4 Individual tool A _ROC values and sensitivity data for HLA-A*0201 using binders and nonbinders gathered from the literature

Tool	A _ROC	0.99 Specificity			0.95 Specificity
		Rank	Sensitivity	Threshold	Rank	Sensitivity	Threshold
Multipred ANN	0.772	11	0.083	5.830	2	0.278	5.600
NetMHC 2.0 ANN	0.772	6	0.139	99.000	7	0.231	300.000
SYFPEITHI	0.762	1	0.194	27.000	2	0.278	25.000
SVMHC MHCPEP	0.745	1	0.194	0.910	4	0.269	0.740
Logistic Regression	0.743	10	0.093	0.424	12	0.167	0.281
ARB Matrix	0.742	8	0.102	1.860	14	0.139	3.550
Bimas	0.722	12	0.074	437.482	9	0.213	159.970
NetMHC 2.0 Matrix	0.719	14	0.046	27.193	10	0.194	23.585
SMM	0.716	4	0.157	100.484	6	0.250	262.434
Rankpep	0.708	3	0.176	89.000	5	0.259	83.000
MHCPred (interactions)	0.707	6	0.139	61.660	1	0.287	100.690
SVMHC SYFPEITHI	0.706	5	0.148	0.970	11	0.185	0.820
HLA Ligand	0.705	12	0.074	147.000	12	0.167	138.000
MHCPred (position only)	0.700	8	0.102	28.250	7	0.231	67.300
Multipred HMM	0.695	16	0.009	7.570	15	0.102	7.290
Predep	0.627	15	0.019	-6.450	16	0.093	-5.330

For details, see the caption for Table 1. The peptides in this literature-derived dataset are available in Additional File 1.