Table 4. Genomic signatures that predict the magnitude of the neutralizing antibody responses using the DAMIP model.

			DAMIP model predictive signatures
			Train on trial 1, test on trial 2										Train on trial 2, test on trial 1
Gene name	Gene symbol	Gene ID	1	2	3	4	5	6	7	8	9	10	1	2	3	4	5
BEN domain-containing 4	BEND4	Hs.120591				×	×	×	×	×		×	×			×
Transcribed locus		Hs.139006	×					×			×	×
6-Phosphofructo-2-kinase/fructose-2,6-biphosphatase 3	PFKFB3	Hs.195471		×
Tumor necrosis factor receptor superfamily, member 17	TNFRSF17	Hs.2556	×	×	×	×	×	×	×	×	×	×	×	×	×	×	×
Tumor protein D52	TPD52	Hs.368433				×			×		×			×		×
Transcribed locus		Hs.481166					×	×		×		×
Kelch repeat and BTB (POZ) domain containing 7	KBTBD7	Hs.63841	×	×	×	×	×	×	×	×	×	×			×		×
Transcribed locus		Hs.649726							×	×	×						×
Nucleosome assembly protein 1-like 2	NAP1L2	Hs.66180			×							×
Accuracy of 10-fold cross-validation (%)			80	80	80	87	87	80	80	80	80	80	89	89	89	89	89
Accuracy of 1-fold blind prediction (%)			100	100	100	100	100	100	100	100	100	100	73	73	73	73	80
Accuracy of 10-fold blind prediction (%)			97	99	94	92	96	98	92	93	93	94	72	71	75	70	79

Analysis of signatures that predict the neutralizing antibody responses. Here all the discriminatory predictive signature sets turned out to consist of day 7 gene expression only. Further, training on trial 1 produces a high blind prediction accuracy on trial 2. TNFRSF17 was present in all the predictive signature sets of the DAMIP model, and several genes, including KBTBD7 and BEND4 appeared in several signature sets.