A novel mutation-proof, next-generation vaccine to fight against upcoming SARS-CoV-2 variants and subvariants, designed through AI enabled approaches and tools, along with the machine learning based immune simulation: A vaccine breakthrough

Abstract

Emerging SARS-CoV-2 variants and subvariants are great concerns for their significant mutations, which are also responsible for vaccine escape. Therefore, the study was undertaken to develop a mutation-proof, next-generation vaccine to protect against all upcoming SARS-CoV-2 variants. We used advanced computational and bioinformatics approaches to develop a multi-epitopic vaccine, especially the AI model for mutation selection and machine learning (ML) strategies for immune simulation. AI enabled and the top-ranked antigenic selection approaches were used to select nine mutations from 835 RBD mutations. We selected twelve common antigenic B cell and T cell epitopes (CTL and HTL) containing the nine RBD mutations and joined them with the adjuvants, PADRE sequence, and suitable linkers. The constructs' binding affinity was confirmed through docking with TLR4/MD2 complex and showed significant binding free energy (-96.67 kcal mol^-1) with positive binding affinity. Similarly, the calculated eigenvalue (2.428517e-05) from the NMA of the complex reveals proper molecular motion and superior residues' flexibility. Immune simulation shows that the candidate can induce a robust immune response. The designed mutation-proof, multi-epitopic vaccine could be a remarkable candidate for upcoming SARS-CoV-2 variants and subvariants. The study method might guide researchers in developing AI-ML and immunoinformatics-based vaccines for infectious disease.

Keywords: Immune simulation; Multi-epitopic peptide vaccine; Mutation-proof; SARS-CoV-2 variants and subvariants.

Fig. 1

Different methodologies were adopted (AI models, machine learning in silico techniques, and immunoinformatics approaches), from mutation selection to designing the next-generation multi-epitopic peptide vaccine candidate of S-protein SARS-CoV-2. (A) A flowchart depicts the methodologies adopted for the mutation selection used in the epitopes for our vaccine design. (B) A 3D model demonstrated all the finally selected nine mutations and their corresponding epitopes in the RBD region of S-protein of different SARS-CoV-2 variants (Alpha, Beta, Delta, Gamma, and Omicron). (C) A flowchart illustrates the methodologies adopted (AI models, machine learning in silico techniques, and immunoinformatics approaches) from mutation selection, multi-epitopic peptide vaccine construct design, and its characterization of several properties such as physicochemical properties, antigenicity, allergenicity, etc.

Fig. 1

Different methodologies were adopted (AI models, machine learning in silico techniques, and immunoinformatics approaches), from mutation selection to designing the next-generation multi-epitopic peptide vaccine candidate of S-protein SARS-CoV-2. (A) A flowchart depicts the methodologies adopted for the mutation selection used in the epitopes for our vaccine design. (B) A 3D model demonstrated all the finally selected nine mutations and their corresponding epitopes in the RBD region of S-protein of different SARS-CoV-2 variants (Alpha, Beta, Delta, Gamma, and Omicron). (C) A flowchart illustrates the methodologies adopted (AI models, machine learning in silico techniques, and immunoinformatics approaches) from mutation selection, multi-epitopic peptide vaccine construct design, and its characterization of several properties such as physicochemical properties, antigenicity, allergenicity, etc.

Fig. 2

The schematic diagram 3D model selected RBD mutations for our vaccine construct. (A) A 3D surface model shows the location of selected RBD mutations within the S-protein. (B) Another ribbon-like model shows the location of selected RBD mutations within the ribbon diagram of the RBD region of the S-protein. (C) A model depicted using a crisscross line diagram demonstrates the relatedness between our selected RBD mutations and the occurrence of SARS-CoV-2 variants. (D) A heatmap was described to understand the positions of the different ranks of our chosen mutations which were used to develop the vaccine construct. Here we used three global ranks (worldwide observed rank, BFE change rank, and antibody disruption rank) retrieved from the mutation analyzer database and depicted the heatmap.

Fig. 3

Population coverage by selected CTL epitopes and their particular MHC binding alleles of designed vaccine construct. We consider two critical alleles for our work (MHC class I and MHC class II) (A) The figure shows the worldwide population coverage of the MHC class I allele, (B) It represents the worldwide population coverage of the MHC class II allele.

Fig. 4

Peptide vaccine construct and its refined 3D structure from the S-proteins of SARS-CoV-2 and its mutational epitopes. (A) The graphical diagram shows the final vaccine construct from antigenic mutational epitopes of S-proteins of SARS-CoV-2 using different peptide linkers such as AAY and GPGPG. (B) Surface structure model. (C) The ribbon model shows the vaccine construct.

Fig. 5

The secondary structure prediction plot of our vaccine construct. The alpha helices were marked in blue color, while extended strands and beta turns were shown in red and green colors, respectively. (A) It demonstrates the visualization of the result of the study and (B) The score curves for each predicted state of peptide chain analysis.

Fig. 6

The secondary structure analysis of the proposed multi-epitopic peptide vaccine using the PESIPRED webserver.

Fig. 7

Structural quality validation of our vaccine construct (A) Structure was validated through the Ramachandran plot for all amino acids residues of the vaccine construct, and it displays the result of the validation (B) Showing the energy plot of amino acids of vaccine candidate (C) The ‘Z’ score of vaccine candidate (black dot) within the ‘Z’ score range of experimentally validated peptide structure.

Fig. 8

The complex structure shows the molecular docked complex. The consequence of our vaccine construct with TLR4/MD2 complex analyzed through Normal Mode Analysis (NMA) (A) Docked complex (peptide vaccine candidate docked with TLR4/MD2 complex) illustrates surface model shown vaccine construct (red), (B) The mobility of the docking complex of vaccine and TLR4/MD2 specified with colored affine-arrow, (C) The deformability plot of the docked complex, (D) The calculated B-factor of NMA and PDB B-factor of the proposed vaccine candidate. (E) The eigenvalue showing for the molecular docking complex, (F) The covariance matrix map of atomic pair of interacting residues (amino acid) of the molecular docked complex, (G) The connection spring map of the elastic network model of the docked complex.

Fig. 9

The specific immune response of human host cells after injection of our designed peptide-based vaccine construct. The study illustrates diverse results from the in silico simulation and machine learning approaches. (A) The simulation graph shows the elevated level of immunoglobulins. Such elevation of immunoglobulins at different concentrations gradients of antigens, (B) It indicates the population of B cells secretory (IgG1, IgG2, and IgM) subsequently three injections of designed vaccine construct (C) The graph showing the analysis consequence of the population per entity-state (presenting on class-II, counts for the active state, internalized the Ag, duplicating antigenic by the diverse color variant). (D) The figure notifies the CTL population in the time (days) gaps after injection of the designed vaccine candidate. (E) The graph illustrates the CTL population in different situations, active and resting in the time (days) gaps after injection of the stated vaccine construct. (F) It depicts the total count (TC) of TH cell populations and the memory cells sub-divided into isotypes of IgG1, IgG2, and IgM after the injection of the developed vaccine. (G) It shows the population count per entity-state of HTL cell count in the active and resting states after the injection of our vaccine, (H) The behavior of the NK cell populations. (I) The graph shows the performance of the DC population in the resting and active states after injection of our designed vaccine. (J) The macrophages cell population after only the peptide vaccination, (K) It shows the concentration of cytokines and interleukins components with Simpson index [D], (L) The graph shows the total count of EC cells that is altered to virus-infected, active, and presenting on MHC class-I molecules.