Reduced amino acid alphabet-based encoding and its impact on modeling influenza antigenic evolution

Currently, vaccination is one of the most efficient ways to control and prevent influenza infection. Vaccine production largely relies on results laboratory assays, including hemagglutination inhibition microneutralization which are time-consuming laborious. Viruses can escape from immune response that in need revise update vaccines biannually. The assay measure how effectively antibodies against a reference strain bind block an antigen test strain. Various computer-aided models have been developed optimize candidate vaccine selection. A general problem modeling antigenic evolution representation genetic sequences for input into research model. Our motivation stems well-known encoding information evolution. This paper introduces two-fold approach based reduced amino acid alphabet index databases called AAindex. We propose apply simplified alphabet, also sub-alphabet or implies use 20 acids being clustered divided groups. proposed allows redefine mutations termed groups located alphabets. investigated 40 sets their performance experimental indicate alphabets achieve standard its accuracy. Moreover, these provide deeper insight various aspects relationship between mutation variation. By checking identified high-impact sites Influenza Research Database, we found not only significant influence antigenicity, but other close proximity. all selected non-antigenic related responses. According experimentally determined be T-cell epitopes, B-cell MHC-binding epitopes different classes. highlighted caveat: while simulating evolution, model should consider sites, neighboring positions, as they may indirectly impact antigenicity. Additionally, our findings structural charge characteristics beneficial agreement with previous studies.