Table 1.

Five-fold cross validation on all training data and CD-HIT filtered training data.

Training Dataset	Alleles	Seq Count	IC 50		Binary Binding
			AUC	SRCC	AUC	SRCC
BD2013	All alleles	121,787	0.94	0.73	0.94	0.70
	HLA-A	72,618	0.94	0.75	0.94	0.73
	HLA-B	46,915	0.94	0.68	0.94	0.64
	HLA-C	2,254	0.89	0.70	0.89	0.69
CD-HIT BD2013	All alleles	104,449	0.94	0.71	0.94	0.68
	HLA-A	60,987	0.94	0.73	0.94	0.71
	HLA-B	41,360	0.94	0.66	0.94	0.62
	HLA-C	2,102	0.89	0.69	0.89	0.68
BD2009	All alleles	88,742	0.93	0.69	0.93	0.68
	HLA-A	57,173	0.93	0.72	0.93	0.71
	HLA-B	31,569	0.93	0.62	0.93	0.60