Table 1.

Summary and examples of computational influenza universal vaccine design.

Approach	Conceptual Design	Evidence-Level	Advantages	Disadvantages	Examples
Consensus-based optimized approach	Figure 2A	Pre-clinical	(1) Efficiently generate a potentially full profile of conserved immunogenicity in viral genome; (2) Induce broad HA inhibition antibody titers that are cross-reactive with diverse strains within the same subtype; (3) Neutralize the receptor binding sites to prevent influenza disease with a clear path towards clinical proof of correlation for protective efficacy in humans	(1) Biased viral samples may not generate consensus sequences that represent full profile of conserved immunogenicity; (2) Large efforts on surveillance data required	Pre-clinical tests on H1, H3 and H5 HA [28,41,42,43,45]
Ancestral sequence reconstruction	Figure 2B	Pre-clinical	(1) Induce broad cross-reactive protection within highly diverse influenza subtype (2) Account for sampling bias and the variability of substitution rates among sites; (3) Potentially avoid the detrimental effects of antigenic drift with ancestral sequences; (4) Incorporate protein functional and structural domains	(1) More sophisticated and advanced models to incorporate protein domains are still under development; (2) Experimental data on protein function is needed	Pre-clinical tests on ancestral sequence of H5N1 HA and NA [44]
Immunomics	Figure 2C	Pre-clinical & Clinical	(1) Account for the heterogeneity of the major histocompatibility complex (MHC) in host; (2) Protections and viral clearance from T-cell response has been distinctively tested	(1) Indirect estimation on epitope affinity to MHC; (2) To keep conformational epitopes to be function when designed into vaccine can be challenge	FP-01.1 Flu-v Multimeric-001 See Table 2 for details