The Impact of Amino Acid Encoding on the Prediction of Antigenic Variants

Determining antigenic variants is a critical task in vaccine production. In recent decades, besides the well-known analysis of hemagglutination inhibition assay, various in silico models have been developed to predict antigenic variants mainly based on the hemagglutinin antigen sequence. One of the initial but necessary steps in modeling is the numerical representation of the protein such that the characteristics of the protein can be reflected more significantly in associated with antigenicity. This paper is devoted to the comparison of various simplified forms of the amino acid alphabet and their performance in predicting the antigenic evolution of the influenza virus. The experimental results indicate that some simplified alphabets are robust than the conventional standard amino acid alphabet in terms of accuracy of predicting antigenic variants of the influenza virus.