COVID-19 coronavirus vaccine design using reverse vaccinology and machine learning

Edison Ong; Mei U Wong; Anthony Huffman; Yongqun He

The multiple sequence alignment and the resulting phylogeny of nsp3 protein showed that 174 this protein in SARS-CoV-2 was more closely related to the human coronaviruses SARS-CoV and 175 MERS-CoV, and bat coronaviruses BtCoV/HKU3, BtCoV/HKU4, and BtCoV/HKU9. We studied 176 the genetic conservation of nsp3 protein ( Figure 1A (Table S4 -5). In terms of linear B cell epitopes, there were 14 epitopes with BepiPred 195 scores over 0.55 and had at least ten amino acids in length (Table S6) (Table 2 ). But the inactivated or attenuated whole virus vaccine might induce strong 218 adverse events. On the other hand, vaccines targeting the structural proteins induce a strong 219 immune response 23,32,33 . In some studies, these structural proteins, including the S and N proteins, 220 were reported to associate with the pathogenesis of coronavirus 24,34 and might raise safety 221 concern 11 . Our study applied state-of-the-art Vaxign reserve vaccinology (RV) and Vaxign-ML 222 only the S protein may induce high serum-neutralizing antibody titers but cannot induce complete 231 protection 10 . In addition, HCoV-NL63 also uses S protein and employs the angiotensin-converting 232 enzyme 2 (ACE2) for cellular entry, despite markedly weak pathogenicity 35 . This suggests that the 233 S protein is not the only factor determining the infection level of a human coronavirus. Thus, 234 alternative vaccine antigens may be considered as potential targets for COVID-19 vaccines. 235