key: cord-0926871-dcpre9wp
authors: Sarkar, Bishajit; Ullah, Md. Asad; Araf, Yusha; Islam, Nafisa Nawal; Zohora, Umme Salma
title: Immunoinformatics-guided designing and in silico analysis of epitope-based polyvalent vaccines against multiple strains of human coronavirus (HCoV)
date: 2021-03-15
journal: Expert review of vaccines
DOI: 10.1080/14760584.2021.1874925
sha: e04cab61f533a1fb3bb77860ddad59cfd9acbf1f
doc_id: 926871
cord_uid: dcpre9wp

Objectives: The group of human coronaviruses (HCoVs) consists of some highly pathogenic viruses that have caused several outbreaks in the past. The newly emerged strain of HCoV, the SARS-CoV-2 is responsible for the recent global pandemic that has already caused the death of hundreds of thousands of people due to the lack of effective therapeutic options. Methods: In this study, immunoinformatics methods were used to design epitope-based polyvalent vaccines which are expected to be effective against four different pathogenic strains of HCoV i.e., HCoV-OC43, HCoV-SARS, HCoV-MERS, and SARS-CoV-2. Results: The constructed vaccines consist of highly antigenic, non-allergenic, nontoxic, conserved, and non-homologous T-cell and B-cell epitopes from all the four viral strains. Therefore, they should be able to provide strong protection against all these strains. Protein-protein docking was performed to predict the best vaccine construct. Later, the MD simulation and immune simulation of the best vaccine construct also predicted satisfactory results. Finally, in silico cloning was performed to develop a mass production strategy of the vaccine. Conclusion: If satisfactory results are achieved in further in vivo and in vitro studies, then the vaccines designed in this study might be effective as preventative measures against the selected HCoV strains.

started in Wuhan, China in December 2019, and as of 20 January 2021, about 9.6 million people from all over the world got infected with the mortality rate of about 2.5% [9] .

Currently, neither any satisfactory antiviral therapy nor any vaccine is available to fight the infections of HCoVs including SARS-CoV, MERS-CoV, and SARS-CoV-2. Different types of antiviral therapies have been tried and experimented over the years but none of these treatments has gained success to the expected threshold. Antiviral preparations like Ribavirin, Lopinavir, and Ritonavir, in combination with various interferons, had been used to alleviate the MERS and SARS viral infections in recent years [7] . However, the results of these studies are inconsistent and due to lack of any definitive proof from the trials, these treatments could not be declared 100% effective to fight the coronaviruses [7, 10] . Although much emphasis has already been put forward to develop potential treatments for COVID-19, still no approved preventing option is available. Several antiviral therapies are being tried by scientists, but these therapies offer varying therapeutic results from patient to patient of COVID-19 [11] . Much progress was reported in vaccine development against the MERS-CoV. However, due to numerous safety and efficacy concerns in the vaccinated individuals, these vaccines have failed to provide any definitive immunity to the viral infections [7, 10] .

In our study, the methods of immunoinformatics and reverse vaccinology have been utilized to design potential epitope-based polyvalent vaccines targeting the spike glycoproteins of multiple strains of HCoV i.e., HCoV-OC43, SARS-CoV, MERS-CoV, and SARS-CoV-2. Reverse vaccinology and immunoinformatics are quick, easy, and costeffective approaches of vaccine designing and development where the novel antigens of a pathogen are identified by analyzing its genetic makeup with the aid of different tools of in silico biology [,] . A flowchart of the methods employed in the vaccine designing in this experiment has been illustrated in Figure 1 . The spike glycoprotein (SG) sequences (also known as S protein) of CoVs are among the most important proteins that mediate the virus entry and pathogenesis, determine the host range of the virus and elicit the effects of the neutralizing antibodies [15] [16] [17] . The SG proteins of HCoVs have two subunits i.e., S1 and S2 subunits. The S1 subunit functions in binding with the host cell receptor and the S2 subunit functions in the fusion of viral and cellular membranes. The S1 subunit has two domains i.e., N-terminal domain (S1-NTD) and the C-terminal domain (S1-CTD). During infection, the S1-CTD subunit recognizes the protein receptors of the host cells like the ACE2, APN, and DPP4 for binding. Then the S2 subunit causes the fusion of the viral and host cell membrane, thus the viral genome is entered into the host cell. Therefore, SG proteins on the surface of HCoVs play very important roles in mediating their infection [18,] . The SG proteins also give the HCoVs their characteristic crown-like appearance because these proteins stud and protrude from the viral envelope. For these reasons, targeting the SGs for designing vaccines against HCoVs presents an effective strategy [20] .

The strains of HCoVs and the target proteins were identified and retrieved from the National Center for Biotechnology Information (NCBI) (https://www.ncbi.nlm.nih.gov/) as well as the UniProt database (https://www.uniprot.org/).

The antigenicity of the target protein sequences was predicted by the online antigenicity prediction tool, VaxiJen v2.0 (http://www.ddg-pharmfac.net/vaxijen/VaxiJen/VaxiJen. htm), keeping the prediction accuracy parameter threshold at 0.4. The accuracy, sensitivity, and specificity of a prediction depend on the accuracy prediction threshold and the 0.4 threshold improves the prediction accuracy of the server [21, 22] . Thereafter, the physicochemical properties i.e., the number of amino acids, theoretical pI, instability indexes, extinction coefficients, half-lives, grand average of hydropathicity (GRAVY) etc. of the target proteins were predicted by ExPASy's online tool ProtParam (https://web. expasy.org/protparam/) [].

The T-cell and B-cell epitopes of the protein sequences were predicted for the vaccine construction. The cytotoxic T-cells are important for specific antigen recognition and the helper T-cells are the most important component of adaptive immunity, which function in activating the B-cell, macrophages, and even cytotoxic T-cells [24, 25] . As a result, both cytotoxic and helper T-cell epitopes are essential for a successful vaccine designing study.

The epitopes were predicted using the Immune Epitope Database or IEDB (https://www.iedb.org/), which contains a vast collection of experimental data on T-cell epitopes and antibodies [26] . The MHC class-I or CD8+ cytotoxic T-lymphocytic (CTL) epitopes were predicted using the IEDB recommended 2020.04 (NetMHCpan EL 4.0) method . The top MHC class-I epitopes were selected for further analysis. Again, MHC class-II or CD4+ helper T-lymphocytic (HTL) epitopes were predicted using the IEDB Recommended Method 2.2 which gives predictions in percentile scores. The top predicted MHC class-II epitopes were considered for further analysis. The determined MHC class-I epitopes were 9-mers and the MHC class-II epitopes were 15-mers. The HLA alleles for which these predicted epitopes were found, were also noted and listed. B-cell epitopes in a vaccine trigger the antigen-specific immunoglobulin production which are crucial components of adaptive immunity []. The B-cell epitopes of the proteins were predicted by BepiPred linear epitope prediction method [] and the epitopes with more than ten amino acids were considered as potential candidates for vaccine construction. Using a combination of a hidden Markov model and a propensity scale method, BepiPred predicts the location of linear B-cell epitopes [].

In this step, the predicted T-cell and B-cell epitopes from the previous step were screened with several tools for predicting their antigenicity, allergenicity, toxicity, conservancy, and human homology. Vaccine components should be highly antigenic and at the same time non-allergenic. Moreover, the vaccine components should also devoid of any toxic reaction. The antigenicity determining tool, VaxiJen v2.0 (http://www. ddg-pharmfac.net/vaxijen/VaxiJen/VaxiJen.htm) was again used in this step where the threshold was kept at 0.4 [21, 22] . Two different tools i.e., AllerTOP v2.0 (https://www.ddgpharm fac.net/AllerTOP/), as well as AllergenFP v1.0 (http://ddgpharmfac.net/AllergenFP/) were used for allergenicity prediction of the epitopes. The AllerTOP v2.0 server has better prediction accuracy of 88.7% than AllergenFP v1.0 server (87.9%) [29, 30] . After that, ToxinPred (http://crdd.osdd.net/raghava/tox inpred/) server was used for toxicity prediction of all the epitopes. The default support vector machine (SVM) method was used for prediction keeping all the parameters default [31] . For conservancy analysis of the epitopes, multiple sequence alignment (MSA) was performed using the online tool Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/ ) as well as the alignment module of UniProt database (https:// www.uniprot.org/). In the MSA, SGs sequences from total 80 different isolates of HCoVOC43, SARS-CoV, MERS-CoV, and SARS-CoV-2 (20 isolates from each of the virus strain) were used (Supplementary Table S1 ). The homology of the epitopes to the human proteome was also determined in this step. The protein BLAST module (blastP) of BLAST (https:// blast.ncbi.nlm.nih.gov/Blast.cgi) tool was used in the human homology determination, where Homo sapiens (taxid: 9606) was used for comparison keeping all other parameters default. An e-value cutoff of 0.05 was set and epitopes that showed no hits below e-value inclusion threshold were selected as nonhomologous pathogen peptides [32] . The epitopes that were found to be highly antigenic, non-allergenic, nontoxic, 100% conserved among all the corresponding isolates and nonhomologous to the human proteome, were considered as the 'best selected epitopes' and used for vaccine construction in the later stages.

The helper T-cells are known to produce different types of cytokines like INF-gamma, IL-4 (interleukin-4), and IL-10 (interleukin-10). These cytokines later aid in the activation of different immune cells i.e., cytotoxic T-cells, macrophages, etc [33] . Therefore, the cytokine inducing capability of the HTL or MHC class-II epitopes were determined in this study. The interferon-gamma (IFNgamma) induction capability of the predicted HTL epitopes was carried out using IFNepitope (http://crdd.osdd.net/raghava/ifnepi tope/) server, using the Hybrid (Motif + SVM) prediction approach. The Hybrid prediction approach is a highly accurate approach for IFN-gamma inducing epitope prediction [34] . Moreover, IL-4 and IL-10 inducing properties of the HTL epitopes were predicted using IL4pred and IL10pred servers, respectively [,36 ]. Both the IL4pred and IL10pred predictions were conducted based on the SVM method where the default threshold values were set at 0.2 and −0.3, respectively. Thereafter, the transmembrane topology experiment of all the epitopes predicted in sub-section 2.3 was carried out using the TMHMM v2.0 server (http://www.cbs.dtu.dk/ services/TMHMM/) [37] .

The distribution of specific HLA alleles among different ethnicities and population around the world is an essential criterion for designing a successful multi-epitope vaccine. The expression of different HLA alleles varies from one ethnicity to another around the world. The IEDB population coverage tool (http://tools.iedb. org/population/) was used to determine the population coverage of the best-selected epitopes across multiple HLA alleles among different nations around the world [38, 39] .

Molecular docking study is one of the necessary steps in reverse vaccinology since it predicts the binding of epitopes with the MHC receptors or sometimes, antibodies [14] . The 3D structures of the best-selected epitopes (epitopes that followed the previously mentioned criteria of high antigenicity, non-allergenicity, nontoxicity, 100% conservancy, and non-homology to human proteome) were taken into consideration for peptide-protein docking. The 3D structure generation process was carried out by PEP-FOLD3 (http:// bioserv.rpbs.univ-paris-diderot.fr/services/PEP-FOLD3/) [40] [41] [42] .

Thereafter, the best-selected MHC class-I epitopes were docked with HLA allele HLA-A*11-01 (PDB ID: 5WJL). And the MHC class-II alleles were docked with HLA DRB1*04-01 allele (PDB ID: 5JLZ). The successful docking of the selected MHC epitopes with the HLA alleles ensured that the epitopes might interact with these MHC alleles during an immune response. PatchDock (https://bioinfo3d.cs.tau.ac.il/PatchDock/php.php) online molecular docking tool was used for the peptideprotein docking. After the docking by the PatchDock tool, the results were refined and rescored by the FireDock server (http://bioinfo3d.cs.tau.ac.il/FireDock/php.php) which provides output in the context of global energies and the lowest global energy represents the best prediction [43] [44] [45] [46] . The docked complexes from the docking experiment were then visualized and analyzed by Discovery Studio Visualizer [47] .

The best-selected epitopes were conjugated with each other to construct three possible vaccines. From these three vaccines, one best possible vaccine would be selected based on the molecular docking analysis. The CTL, HTL and BCL epitopes were conjugated by GGGGS, GPGPG, and KK linkers. Three vaccine constructs contained three different adjuvant sequences: L7/L12 ribosomal protein and HBHA protein (M. tuberculosis, accession number: AGV15514.1) and betadefensin-3 (UniProt accession number: Q5U7J2), that were linked to the epitopes by EAAAK linkers. Adjuvants are known to enhance the antigenicity, immunogenicity, stability, and longevity of the constructed vaccines [48, 49] . The vaccine constructs also contained the pan HLA-DR epitope (PADRE) sequence, attached with the adjuvant and epitopes. The EAAAK linkers provideeffective separation of domains of the bifunctional fusion proteins [50] . Again, the GPGPG linkers have the capability to prevent the generation of junctional epitopes and facilitate the immune processing and presentation [51] . Furthermore, the bi-lysine (KK) linkers preserve the independent immunological activities of the epitopes of a vaccine [52] . And the GGGGS is a flexible linker which was proved to be effective in conferring resistance to proteases [53] .

Significant elevation of toll-like receptor-2 (TLR-2) and TLR-4 was reported in different studies after the HCoV infections in human body [,54] . L7/L12 ribosomal protein and HABA protein can stimulate the activity of TLR-4 and beta-defensin has the capability to activate TLR-1, 2, 4, etc. Moreover, studies have also shown that the PADRE sequence improves the immune response enhancing ability of the CTL epitopes. It also increases the potency of vaccines with minimal toxicity. For this reason, the adjuvants and PADRE sequence might confer potential immunity against the HCoV infections [55] [56] [57] [58] [59] [60] [61] [62] . Figure 2 represents a schematic diagram of the three vaccines in their appropriate orientation that were constructed using different adjuvants and linkers. The three vaccines differ from each other only in their adjuvant sequences.

The antigenicity of the three vaccine constructs was predicted by the VaxiJen v2.0 (http://www.ddg-pharmfac.net/vaxijen/ VaxiJen/VaxiJen.htm) keeping the threshold at 0.4 [21, 22] . The predicted results of the Vaxijen v2.0 server were further cross-checked by the ANTIGENpro module of the SCRATCH protein predictor (http://scratch.proteomics.ics.uci.edu/) [63] . The constructed vaccines should be highly antigenic to stimulate a better immune response. The allergenicity of the vaccine constructs was predicted by three different online tools i.e., AlgPred (http://crdd.osdd.net/raghava/algpred/), AllerTop v2.0 (https://www.ddgpharmfac.net/AllerTOP/), and AllergenFP v1.0 (http://ddg-pharmfac.net/AllergenFP/) for improving the prediction accuracy. The AlgPred (http://crdd.osdd.net/ raghava/algpred/) server attempts to integrate different approaches of allergenicity determination to predict potential allergenic proteins with high accuracy [64] . Thereafter, the physicochemical properties of the constructed vaccines were predicted again by the online tool ProtParam (https://web. expasy.org/protparam/) []. Along with the physicochemical property analysis, the solubility of vaccine constructs upon over-expression in E. coli host was predicted by the SOLpro module of the SCRATCH protein predictor (http://scratch.pro teomics.ics.uci.edu/) and later further clarified by the Proteinsol server (https://protein-sol.manchester.ac.uk/) [,66].

The secondary structures of the vaccine constructs were predicted by several online tools for improving prediction accuracy i.e., PRISPRED (http://bioinf.cs.ucl.ac.uk/psipred/) (using PRISPRED 4.0 prediction method), GOR IV (https://npsa-prabi. ibcp.fr/cgibin/npsa_automat.pl?page=/NPSA/npsa_gor4.html), SOPMA (https://npsa-prabi.ibcp.fr/cgibin/npsa_automat.pl? page=/NPSA/npsa_sopma.html), and SIMPA96 (https://npsaprabi.ibcp.fr/cgi-bin/npsa_automat.pl?page=/NPSA/npsa_ simpa96.html). These tools are easy, accurate and efficient online servers to predict the amounts or percentage of amino acids in αhelix, β-sheet and coil structure formations [67] [68] [69] [70] [71] [72] . All the parameters were kept default while predicting the secondary structures of the vaccine constructs. On the other hand, RaptorX (http://raptorx.uchicago.edu/) tool was used for predicting the tertiary or 3D structures of the three vaccine constructs. The server predicts the tertiary structure of a query protein or peptide by template-based method [73] [74] [75] .

When protein 3D structures are predicted by computational methods, they may lack their true, native structures. Therefore, the 3D structure refinement was conducted to convert the low resolution predicted model to models that closely resemble the native protein structure. The generated 3D structures of the vaccines were refined by GalaxyRefine module of the GalaxyWEB server (http://galaxy.seoklab.org/). The tool uses CASP10 tested refinement method and dynamics simulation to provide better-refined structures [76] [77] [78] .

After structure refinement, the vaccine constructs were validated by analyzing the Ramachandran plots, generated by PROCHECK (https://servicesn.mbi.ucla.edu/PROCHECK/) server [79, 80] . Moreover, another online tool, ProSA-web (https:// prosa.services.came.sbg.ac.at/prosa.php) was also used for protein validation. It is a tool for checking the 3D models of protein structures for potential errors. The server generates z-score which is used to express the quality of a query protein structure. A z-score within the range of the z-scores of all the experimentally determined protein chains in the current PDB database represents better quality of a query protein [81] .

Since disulfide bonds are necessary for the conformational stability of folded proteins, the disulfide engineering of the vaccine constructs was carried out in this experiment. The disulfide bond prediction was conducted by the Disulfide byDesign 2 v12.2 (http://cptweb.cpt.wayne.edu/DbD2/) server [82] . During the experiment, the χ3 angle was kept −87° or +97° ±5 to discard numerous putative disulfides which were generated using the default angles of +97° ±30° and −87° ±30°. The mentioned χ3 angle was used for identifying the fewer amount of disulfide bonds with desired characteristics. Again, the Cα-Cβ-Sγ angle was set at its default value (114.6° ±10) because studies have estimated that the Cα-Cβ-Sγ angle reaches a peak near 115° and covers a range from 105° to 125° in known disulfide bonds. Finally, the residue pairs with energy less than 2.2 Kcal/mol were selected and mutated to cysteine residue to allow disulfide bridge formation [83] . The energy value 2.2 Kcal/mol was selected as threshold because 90% of native disulfide bonds are generally found to have energy value less than 2.2 Kcal/ mol [82] .

In protein-protein docking, the constructed vaccines were docked against multiple MHC alleles and TLRs and one best vaccine was selected based on their docking scores. The vaccines should have a good binding affinity with the MHC protein segments encoded by different MHC alleles. This is important because the MHC molecules produce potential immune responses after recognizing the vaccines which mimic the original viral infections. Thus vaccines help to produce immunity toward a particular virus or viruses [84] . In this study, the vaccine constructs were docked against the MHC alleles i.e., HLA-A*01:01 (PDB ID: 4NQX), HLA-A*02:01 (PDB ID: 4U6X), HLA-A*03:01 (PDB ID: 2XPG), and HLA-A*11:01 (PDB ID: 5WJL). Moreover, TLR-8 is responsible for generating immune responses against RNA viruses and TLR-3 aids in immune response generation against the DNA viruses [85, 86] . Since HCoVs are positive-sense RNA viruses, the vaccines were also docked against the TLR-8 (PDB ID: 3W3M) []. To further assess the efficacy of the vaccines, they were docked with two more TLRs i.e., TLR-2 (PDB ID: 6NIG), and TLR-4 (PDB ID: 4G8A) because TLRs are some of the main players of the immune system.

In order to improve the prediction accuracy, the proteinprotein docking was carried out by three different online tools. At first, the docking was carried out by ClusPro 2.0 (https:// cluspro.bu.edu/login.php) where the lower energy score corresponds to the better binding affinity [88] [89] [90] . The ClusPro server calculates the energy score based on the following equation:

Thereafter,thedockingwasagaincarriedoutbythePatchDock (https://bioinfo3d.cs.tau.ac.il/PatchDock/php.php) server and the results were then refined by FireDock server (http://bioin fo3d.cs.tau.ac.il/FireDock/php.php) [43, 44, 46] . Finally, another round of docking was performed by the HawkDock server (http://cadd.zju.edu.cn/hawkdock/) along with the Molecular Mechanics/Generalized Born Surface Area (MM-GBSA) study. In these servers, the lower output scores correspond to better binding affinity and vice versa [91] [92] [93] [94] . Visualization of the best vaccine-TLR-8 complex was carried out by Discovery Studio Visualizer [47] . The vaccine showing the best performance in the docking study was considered as the bestselected/predicted vaccine construct.

Antibody-mediated humoral immunity initiates within the body when the B-cells interact with their epitopic counterparts. Therefore, the vaccines should have effective conformational B-cell epitopes to provide better immunity. The conformational B-cell epitopes of the best-predicted vaccine protein from the 2.13 sub-section were determined by IEDB ElliPro tool (http://tools.iedb.org/ellipro/), using the default parameters of a minimum score of 0.5 and a maximum distance of 6 angstrom [95].

The molecular dynamics simulation study was conducted by the online simulation tool, iMODS (http://imods.chaconlab.org/). It is a fast, easy and user-friendly tool for dynamics simulation study. The dynamics simulation was conducted only for the complex containing the best-selected vaccine (selected based on the docking study) and TLR-8. Different dynamic parameters, for example, deformability, B-factor (mobility profiles), eigenvalues, variance, co-variance map, and elastic network of the protein complex are predicted quite efficiently by this server. The deformability of a protein can be defined as the ability of the amino acids to deform at a specific position within the protein backbone. Moreover, the deformability is also depended on the eigenvalue of a protein.

Lower eigenvalue represents easy deformability of the protein complex. The dynamics simulation study was conducted to predict the stability of the vaccine construct [96-100].

The immune simulation study was conducted for the bestpredicted vaccine to predict its immunogenicity and immune response profile. The C-ImmSim server (http://150.146.2.1/ CIMMSIM/index.php) was used for the immune simulation study which predicts the real-life-like immune interactions using position-specific scoring matrix (PSSM) and machine learning techniques [101] . During the experiment, all the parameters were kept default, except the time steps which were set at 1, 84, and 170 (time step 1 is injection at time = 0 and each time step is 8 hours) . So, three injections are expected to be required at four weeks interval because the recommended interval between two doses of most of the commercial vaccines is four weeks [102] . The Simpson's Diversity index, D was calculated from the figures.

Codon adaptation and in silico cloning are two important steps in reverse vaccinology. An amino acid can be encoded by more than one codon in different organisms, a phenomenon known as codon bias. The cellular machinery of an organism may be completely different from another organism, so the same amino acid can be encoded by completely different codons in different organisms. For this reason, codon adaptation was conducted in this study to predict the suitable codon that might encode a specific amino acid efficiently in a particular organism. In codon adaptation study, the best-predicted vaccine was reverse translated to the possible DNA sequence that is expected to encode the vaccine protein [103, 104] . The Java Codon Adaptation Tool or JCat server (http://www.jcat.de/) was used for the codon adaptation study which ensures maximal expression of a protein in a target organism [105] . Prokaryotic E. coli strain K12 was selected as the target organism and rhoindependent transcription terminators, prokaryotic ribosome binding sites and EaeI and StyI cleavage sites of restriction enzymes, were avoided at the server. The vaccine protein sequence was reverse translated to the optimized DNA sequence by the server. Then the optimized DNA sequence was collected and EaeI and StyI restriction sites were attached to the N-terminal and C-terminal sites, respectively. Finally, the SnapGene restriction cloning software was used for inserting the newly adapted DNA sequence between the EaeI and StyI restriction sites of the pETite vector (Lucigen, USA) [106] . The pETite vector plasmid contains a small ubiquitin-like modifier or SUMO-tag and 6X-His tag which facilitates the solubilization and effective affinity purification of the recombinant protein [107] .

The mRNA secondary structure prediction was conducted by two online servers i.e., Mfold (http://unafold.rna.albany.edu/?q= mfold) and RNAfold (http://rna.tbi.univie.ac.at/cgibin/ RNAWebSuite/RNAfold.cgi). Both of these servers predict the mRNA secondary structures thermodynamically and provide minimum free energy (ΔG Kcal/mol) for each of the generated structures. The lower the minimum free energy, the more stable the folded mRNA is and vice versa [108] [109] [110] [111] . To predict the mRNA folding and secondary structure of the best-selected vaccine, at first the optimized DNA sequence from JCat server was taken and converted to possible RNA sequence by DNA<->RNA>Protein tool (http://biomodel.uah.es/en/lab/cybertory/analysis/trans.htm). Then the RNA sequence was collected from the tool and pasted in the Mfold and RNAfold servers for prediction using the default settings for all the parameters.

The SG sequences of HCoV-OC43, SARS-CoV, MERS-CoV, and SARS-CoV-2 were identified as potential targets for the current study by reviewing the literatures from the NCBI database. The protein sequences of HCoV-OC43, SARS-CoV, and MERS-CoV were retrieved from UniProt and the protein sequence of SARS-CoV-2 was retrieved from NCBI database ( Table 1 ).

All the SG sequences from the four viral strains were found to be antigenic. All of them had a predicted half-life of 30 h in mammalian reticulocytes and all of them were predicted to be stable. The SG from HCoV-OC43 had the highest predicted aliphatic index of 85.94 as well as the highest predicted extinction co-efficient of 196,595 M −1 cm −1 and the SG of SARS-CoV-2 was found to have the lowest GRAVY value of −0.079. Except the SARS-CoV-2, all the other proteins from the selected viruses had almost similar theoretical pI (Supplementary Table S2 ).

The T-cell and B-cell epitopes of the SG sequences of the four viral strains were predicted for vaccine construction. These epitopes were expected to stimulate potential T-cell and B-cell immune responses. Based on their rankings, the top MHC class-I and MHC class-II epitopes as well as B-cell epitopes with length over ten amino acids, were taken into consideration for all the viral strains. From these epitopes, the highly antigenic, non-allergenic, nontoxic, 100% conserved and non-homologous epitopes were considered as the best selected epitopes and selected for final vaccine construction. IFN-gamma, IL-4, and IL-10 inducing capacity prediction of the HTL epitopes had showed that most of the selected HTL epitopes had at least one of these cytokine producing capabilities. Moreover, about 30% of all the epitopes were found to reside outside of the cell membrane according to the transmembrane topology test. Supplementary Table S3, Supplementary Table S4,  Supplementary Table S5 and Supplementary Table S6 list the epitopes predicted for the four viral strains.

The antigenicity and allergenicity, toxicity, conservancy, and human homology predictions were carried out to find out the best epitopes that followed the previously mentioned selection criteria which could be used effectively in vaccine construction. Most of the epitopes were found to be 100% conserved among the SG sequences of the selected isolates and almost all of them were also found to be non-homologous to the human proteome. The best-selected epitopes of the selected virus strains are listed in Table 2 .

The population coverage analysis showed that the best-selected MHC class-I and MHC class-II epitopes covered about 86.11% and 84.94% of the world population, respectively and the epitopes covered 91.23% of the world population in combination. The highest prevalence of the MHC class-I epitopes was found in China (90.49%) and the lowest prevalence of the epitopes were found in Central Africa (54.29%). India was predicted to have 91.40% of the population covered by the MHC class-II epitopes and only 43.67% of the West African population was predicted to have the best-selected MHC class-II coverage. The highest combining coverage of MHC class-I and ClassII epitopes was predicted to be in Saudi Arabia (95.43%) and lowest in West Africa (51.90%). The results of the population coverage analysis among some HCoV infected countries are visualized in Supplementary Figure S1.

All the best-selected T-cell epitopes were subjected to the PEP-FOLD3 server for 3D structure generation. Thereafter, the 3D structures were used in the peptide-protein docking experiment to determine, whether all of the best-selected epitopes had the ability Figure S2 & S3 illustrate the docked epitopes within the binding pocket of their respective targets. Since all the best-selected epitopes were docked successfully with their receptors, it can be concluded that all of the receptors have good affinities toward the MHC class-I and MHC class-II epitopes.

After successful docking, the best-selected epitopes were joined with one another by different types of linkers to construct the vaccine. GGGGS linkers were used to conjugate the CTL epitopes, GPGPG linkers were used to join HTL epitopes and KK linkers were used to attach the BCL epitopes. The PADRE sequence and the adjuvants were linked by EAAAK linkers. The three vaccines were designated as: V1, V2, and V3 ( Table 3 ). The three vaccines differ from each other only in their adjuvant sequences.

All the three vaccine constructs were found to be potentially antigenic and non-allergenic. Moreover, all of them had quite similar theoretical pI of more than 9.00, so all of them were predicted to be basic in nature. However, V2 had the highest predicted extinction co-efficient of 91,260 M −1 cm −1 and it also had the lowest GRAVY value of −0.403. Furthermore, all the three vaccine constructs were found to be stable and had similar predicted half-lives and all of them were found to be soluble upon over-expression in E. coli. The results of the physicochemical property analysis are listed in Table 4 .

The secondary structure prediction of the three vaccine constructs were carried out by several online tools for improving the accuracy of the prediction. The results of the secondary structure prediction are listed in Supplementary Table S8 . All the proteins were found to have quite similar predictions by all the servers. The secondary structure prediction showed that the adjuvants had caused potential differences among the secondary structures of the vaccine proteins. Supplementary Figure S4 illustrates the results of secondary structure prediction carried out by the PRISPRED server. The tertiary structures of the three vaccine constructs were conducted by the RaptorX server. Supplementary Figure S5 illustrates the 3D structures generated by the RaptorX server.

The tertiary structures of the vaccine constructs were refined by GalaxyRefine module of the GalaxyWEB server and later validated by analyzing the Ramachandran plots and z-scores. The protein structure validation of the three vaccine constructs showed that all the three vaccines were much more improved than their crude, unrefined structures. The V1, V2, and V3 had z-scores of 6.87, −5.67, and −5.73, respectively, which reflected that all of them had scores well within the range of experimentally proven X-ray crystal structures [81] . Again, all of them had almost similar percentages of amino acids in the Rama favored region (Supplementary Table S9 ). Figure S6 depicts the Ramachandran plot of each of the three vaccine constructs (generated by the PROCHECK server) and z-score or model quality graphs generated by the ProSA-web server.

The V1, V2, and V3 vaccine constructs generated 23 Figure S7 gives the visual representation of the selected amino acid pairs in their original form and mutated form for disulfide bond formation. With 8 pairs of possible disulfide bonds, V2 was found to be most stable among the three vaccine constructs.

Much emphasis was given on molecular docking analysis of the three vaccine constructs with several MHC alleles and TLR-8 because one best vaccine construct was predicted based on the docking analysis. Therefore, to improve the prediction accuracy, the docking was carried out by three different online tools. From the docking analysis listed in Table 5 , it can be declared that V3 vaccine construct was the best vaccine construct. It generated the lowest and so the best binding free energies as well as ClusPro global energy scores with all the target proteins in the MM-GBSA and ClusPro analyses, respectively. Again, V3 also generated the best scores with all the targets except the HLA-A*01:01 allele, when docked by HawkDock server. Therefore, this server also pointed toward V3 as the best vaccine construct. Moreover, when analyzed by PatchDock and FireDock servers, V3 was predicted to have best scores with HLA-A*01:01, HLA-A*11:01, TLR-2, TLR-4, and TLR-8. And most importantly, V3 showed the satisfactory performances when docked against the TLR-2, TLR-4, and TLR-8 by all the servers. None of the V1 and V2 vaccine constructs generated such good results like V3. For this reason, V3 vaccine construct was considered to be the best vaccine construct and the later analyses were conducted only for the V3 construct. Figure 3 illustrates the interaction of the best vaccine V3 and the receptor protein TLR-8 and their interacting amino acids.

The conformational B-cell epitopes were predicted using ElliPro server which predicts conformational epitopes from tertiary structures. Total 357 residues were found with scores varying from 0.516 to 0.876. The length of the epitopes were ranged from 3 to 158 amino acids and predicted to be located within five conformational B-cell epitopes ( Supplementary  Table S10 and Supplementary Figure S8 ). Figure 4 illustrates the results of molecular dynamics simulation of V3-TLR-8 docked complex. The molecular dynamics simulation study is performed in revere vaccinology to determine the stability and physical movements of atoms and molecules of a vaccine construct [112] . Therefore, the simulation study was conducted in this study to determine the movements of molecules in the V3 vaccine construct. The deformability graph of the complex illustrates the peaks in the graphs which represent the regions of the protein with deformability (Figure 4b) . The B-factor graph of the docked complex provides an easy understanding and visualization of the comparison between the NMA and PDB field of the complex (Figure 4c ). The graph of eigenvalue of the complex is illustrated in Figure 4d which shows that V3 and TLR-8 docked complex generated an eigenvalue of 1.278806e-07. The variance graph represents the individual variance by red colored bars and cumulative variance by green colored bars (Figure 4e ). Figure 4f illustrates the co-variance map of the complex where the correlated motion between a pair of residues is depicted by red color, uncorrelated motion is represented by white color, and anticorrelated motion is indicated by blue color. Moreover, the elastic map of the complex represents the connection between the atoms and darker gray regions indicate stiffer regions ( Figure  4g ) [98] [99] [100] .

The immune simulation of the best-selected vaccine V3 was performed by the C-ImmSimm server which predicts the generation of adaptive immunity as well as epitopes and immune interactions [101] . The simulation study had revealed that after each of the three vaccine injections, the primary immune response against the antigenic fragments was predicted to increase 5j and figure 5k ). The vaccine was also able to generate good amount of various types of cytokines like the IFN-gamma, IL-23, IL-10, and IFN-beta which are the most significant cytokines for generating immune response against viruses (Figure 5l) . Overall, the immune simulation study had showed that with the predicted capability of generating high levels of immunoglobulins, active B-cells and T-cell, cytokines and APCs, the polyvalent vaccine V3 might be able to provide good immunogenic protection against the targeted four HCoVs.

The codon adaptation experiment of best vaccine construct V3 revealed that the adapted sequence had Codon Adaptation Index of 0.94 and GC content of 52.4%. The newly adapted DNA sequence was then inserted into EaeI and StyI restriction sites of the pETite vector plasmid. Upon translation in the E. coli host, the vaccine protein was expected to be expressed in fusion with SUMO protein and 6X His tag, which should help effective purification and solubilization of the protein [107] (Supplementary Figure 9 and Figure 6 ). The newly constructed plasmid with the V3 sequence was designated as 'CSMV_3' plasmid. When the lowest minimum free energy of the V3 vaccine was determined by the Mfold server, the best predicted structure (among 46 generated structures by the server) for the optimized construct showed ΔG value of −656.30 kcal/mol. The outcomes of the Mfold server were in agreement with the data generated by RNAfold server, where the ΔG value of the mRNA structure was predicted to be −617.70 kcal/mol. Supplementary Figure S10 illustrates the predicted secondary structure of the V3 vaccine.

Vaccines are widely produced and used worldwide to control and prevent the infections caused by many types of pathogens. Conventional approaches are mainly used for vaccine development and production, although they are costly and time-consuming [14] . In contrast to such conventional methods of vaccine development, today's cutting-edge research and technology as well as the availability of information about the genome and proteome of almost all the viruses and organisms has made it possible to design and develop novel peptide-based 'subunit vaccines' comprised of the antigenic protein portions from a target pathogen. A major advantage of subunit vaccines is that toxic and immunogenic parts of an antigen can be eliminated during a vaccine designing study so that the vaccine would be safe to use in humans [113] [114] [115] . This had led to the development of techniques of bioinformatics and immunoinformatics which can be explored to design novel subunit vaccines in a safe, effective, efficient and inexpensive way [116, 117] . In this study, these methods of immunoinformatics were exploited to design possible polyvalent vaccines against multiple strains of the HCoVs i.e., HCoV-OC43, SARS-CoV, MERS-CoV, and SARS-CoV-2.

The target protein sequences, spike glycoproteins (SGs) of the four HCoV strains, were identified and retrieved from the NCBI database. Thereafter, their antigenicity and physicochemical properties i.e., theoretical pI, extinction co-efficient, estimated half-life, instability index, aliphatic index, grand average of hydropathicity (GRAVY), and molecular weight were determined. All the proteins were found to be antigenic. The theoretical pI describes the pH at which aprotein should have no net charge. The extinction co-efficient of acompound depicts the amount of light that is absorbed by that compound at acertain wavelength [118, 119] . The instability index of acompound describes the probability of that particular compound to be stable and acompound with instability an index over 40 is considered to be unstable [120] . The aliphatic index of aprotein refers to the relative volume of the amino acids in its side chains occupied by the aliphatic amino acids i. e., alanine, valine, etc [121] . The GRAVY value of aprotein is represented as the sum of hydropathy values of all the amino acids of that protein which is then divided by the total number of residues in the protein sequence. The negative GRAVY value represents hydrophilic characteristic and the positive GRAVY value represents hydrophobic characteristic of acompound [122, 123] . All the SG proteins from the four viruses were found to be potentially antigenic. The highest extinction co-efficient of SG from HCoV-OC43 of 196,595 M −1 cm −1 describes that it was predicted to absorb the highest amount of light at acertain wavelength. Again, since all of the proteins had instability index of less than 40, all of them were predicted to be stable and have the same amount of aliphatic amino acids in their side chains, according to the aliphatic index prediction. Moreover, all of the proteins had theoretical halflife of 30 h and their negative GRAVY values represented their hydrophilic characteristics.

An effective multi-epitope subunit vaccine should contain the CTL, HTL, and B-cell epitopes so that during the immunogenic response, the vaccine will stimulate the cytotoxic T-cells, helper Tcells, and B-cell. These cells are the most important cells that provide immunity to the body [115] . The B-cells mediate the humoral immune response by producing antibodies and keeping the memory of a previous infection. However, the humoral immune response provided by the Bcells may get weaker overtime and pathogens or antigens can easily overcome the humoral immune response [124] . In such cases, the cell-mediated immune response can provide much broader, often life-long immunity by secreting antiviral cytokines and specifically identifying and destroying the infected cells. For this reason, the T-cell and B-cell epitopes were predicted for the vaccine construction.

The possible T-cell and B-cell epitopes of the proteins were predicted for vaccine construction so that the vaccines would be able to provoke potential immune responses. However, since a lot of possible epitopes were generated by the IEDB server, some special criteria were set for final selection of the best epitopes for vaccine construction. The epitopes must be highly antigenic because if they are not antigenic, then they would not be able to induce strong immune response. Again, the epitopes have to be non-allergenic and nontoxic so that they would not be able to produce any harmful, toxic and allergenic reaction within the body. Again, since polyvalent vaccines were designed in this study, so the epitopes that were found to be 100% conserved among the selected SG proteins of different isolates, were considered as potential epitopes for the polyvalent vaccines. The 100% conservancy had ensured their efficacy and potency among different isolates of the HCoV viruses. Furthermore, the epitopes must be non-homologous to the human proteome because if they are homologous, then they won't be recognized as foreign antigenic sequences or particles. The epitopes that followed these criteria were considered as the best-selected epitopes and used for vaccine construction. Again, since cytokines like the IFNgamma, IL-10, and IL-4 are required for the activation of many immune cells [33] , the cytokine production ability of the HTL epitopes was also predicted. Most of the HTL epitopes were found to be at least one cytokine inducer (among IFNgamma, IL-10, and IL-4) and all the HTL epitopes selected for vaccine construction were also found to possess at least on cytokine inducing ability. This ability would impact greatly on the immunogenic activities of the vaccine. In the next step, the population coverage analysis was conducted. The population coverage analysis of the MHC epitopes and alleles showed that significant portions of the population around different countries of the world possess the epitopes and respective alleles within their genome. The prevalence of the selected alleles and epitopes was relatively higher among the Asian population. Then the molecular docking study of the best-selected epitopes was carried out. Since all the epitopes showed good results in the docking study, it can be concluded that all of the best-selected epitopes had the capability to bind with their respective MHC class-I and MHC class-II molecules. After successful docking, the best-selected epitopes were conjugated with each other using the appropriate linkers (EAAAK, GGGGS, GPGPG, KK) to construct three vaccines, V1, V2, and V3. All these vaccines were predicted to be highly antigenic as well as nonallergenic. For this reason, they might provoke high immune responses and at the same time, they would not cause any unwanted allergenic reactions within the body.

Thereafter, the physicochemical properties of the three vaccine constructs were determined. All the proteins were found to be basic with theoretical pI of more than 9.00 and quite stable with very good instability indexes (less than 40). Again, the aliphatic index represents the protein's thermal stability and the higher the aliphatic index of a protein, the more thermostable it is. Since all the vaccine constructs were predicted to have quite high aliphatic indexes, all of them were considered to be quite thermostable. Furthermore, the negative GRAVY value of the vaccine constructs revealed that all of them might be hydrophilic in nature. Moreover, all of the vaccine constructs were predicted to have more than 10 h of half-life in E. coli, which might not be a problem during the mass production of the vaccines. Solubility is one of the major factors for the post-production studies. The more soluble a protein on overexpression, the easier the purification of that protein is []. Since all the proteins were predicted to be soluble upon over-expression in E. coli by both servers (SolPro and Protein-Sol), therefore, their purification steps should be much easier. Considering all these aspects, the predicted vaccine constructs might be suitable for a potential vaccine candidate. Figure 6 . The results of the in silico cloning study of the V3 vaccine construct. The adapted DNA sequence of the V3 vaccine was inserted into the pETite plasmid between EaeI and StyI restriction sites which is indicated by the red color in the plasmid. The plasmid with the sequence had been designated as 'CSMV_3' plasmid.

The secondary and tertiary structure prediction of the three vaccine constructs showed that the adjuvant sequences had caused some significant differences among the three vaccines. All the tools predicted almost similar percentages of amino acids in the α-helix, β-strand, and coil structure formation for each of the vaccines. After secondary structure determination, the tertiary structures of the three vaccine constructs were predicted. The tertiary structure refinement and validation study revealed that the quality of all the three vaccine constructs were significantly improved after refinement in the context of GDT-HA, MolProbity, Rama favored amino acid percentage and zscores. All the three refined structures of the vaccine constructs resembled the crystallographic structures and all of them had quite satisfactory Rama favored amino acid percentage with very few amino acids in the outlier regions. After that, the refined structures were used for vaccine protein disulfide engineering. Since disulfide bonds confer structural stability to the proteins, V2 construct with its eight disulfide bonds was predicted to be more stable than the other two constructs.

The docking of the three vaccine constructs with different MHC alleles as well as TLR-2, TLR-4, and TLR-8 was conducted to find out the best vaccine construct. The docking step is very important for two reasons: it would help to predict the interaction ability of the vaccines with different MHC alleles and the docking would also aid in prediction of the best vaccine from these three constructs. So, the docking was conducted using several online tools for better prediction accuracy. When analyzing the docking results, all the three vaccine constructs showed good capability of binding to all the MHC alleles and also the TLRs. However, V3 was considered as the best vaccine construct since the MM-GBSA study and docking studies of V3 by all the servers predicted the best scores against almost all the targets. Although, V1 and V2 also showed quite good outcomes in some aspects, the best results were generated by V3 construct with most of the receptor proteins. And V3 also showed the best docking results by all the servers when docked with the selected TLRs. Therefore, V3 was considered as the best vaccine construct and selected for further analysis.

For providing a strong humoral immunity, a vaccine construct should have conformational B-cell epitopes because these epitopes would aid to activate and stimulate the B-cells when these cells encounter them. Therefore, the discontinuous conformational B-cell epitopes on the surface of the best vaccine V3 were predicted that can mimic the original infection and thus stimulate the antibody production [95] . Total 357 amino acids in five different B-cell epitopic regions were identified with scores ranging from 0.516 to 0.876. The molecular dynamics simulation study was performed for docked TLR-8 and V3 complex using the online tool iMODS. The study showed that the complex had less chance of deformability with quite high eigenvalue of 1.278806e-07. The deformability graph (Figure 4b ) had confirmed that the location of the hinges in the complex were not quite significant and therefore, the complex might have good stability with a lower degree of deforming for each individual amino acid residue. Moreover, the complex also had a good number of correlated amino acids and a large number of stiffer regions. Therefore, the V3-TLR-8 complex showed good results in the molecular dynamics simulation study.

The immune simulation study of the best-selected vaccine V3 showed that the immune response of the vaccine within the host might be consistent with the typical immune response. After each of the vaccine injections, the primary immune response was found to be triggered, which later activated the secondary immune response. An increase in the concentrations of the memory Bcells, plasma B-cell, cytotoxic T-cells, and helper T-cells as well as different antibodies indicated that a good humoral and cell-mediated immune response might have been built in the body after each of the vaccinations and the memory B-cell was predicted to last for several months after the vaccinations. Again, the stimulated helper T-cells will also aid in increasing the growth and proliferation of B-cells, thus elevating the adaptive immunity. Moreover, the increase in the concentrations of macrophages and dendritic cells ensured very good antigen presentation and the increase in the cytokine profile after each of the vaccinations might also contribute to the immunity provided by the vaccine. On the other hand, the negligible Simpson index (D) suggests a diverse immune response [101] . Since the vaccine construct contained multiple numbers of B and T-cell epitopes, so it would be able to generate a diverse immune response. As a result, it can be declared that the vaccine V3 might be able to generate good immune response in the body.

Finally, codon adaptation and in silico cloning studies were carried out to identify the possible codons for expressing the vaccine V3 in E. coli strain K12. The in silico cloning was conducted for E. coli as the host organism because E. coli expression is the recommended system for the production of recombinant proteins. During codon adaptation, CAI value 0.94 and GC content of 52.4% was generated. The optimal range and limit of the CAI value were measured to be close to 1.0 but the any value greater than 0.80 can be considered as good score. Again, the optimal range for GC content of an optimized DNA sequence should be 30% to 70% [103. Therefore, the CAI value of 0.94 and GC content of 52.4% can be declared as very good scores. Thereafter, the optimized sequence was inserted into the EaeI and StyI restriction sites of pETite plasmid vector so that the efficient expression of the vaccine protein would occur. Moreover, the expressed protein might have SUMO and 6X His tag fused with it since the pETite plasmid offers these tags as fusion partners. Fusion of these tags would help the downstream processing of the vaccine. The newly constructed plasmid with the vaccine V3 sequence was designated as 'CSMV_3' plasmid. When the stability of the mRNA secondary structure of the vaccine protein was determined, the both Mfold and RNAfold servers generated negative and much lower minimal free energies of −656.30 and −617.70 kcal/mol, respectively. Since, the lower minimal free energy always represents better mRNA stability, so it can be concluded that the predicted vaccine protein might be quite stable upon transcription.

The genome based technology will continue to dominate in the field of vaccine development. The current COVID-19 pandemic will require much more attention from the immunoinformatics field to design and develop potential vaccine candidates which will be effective to combat the SARS-CoV-2. Scientists from around the world have already started working on designing vaccine using this novel approach. Yazdani et al., Joshi et al., Lizbeth et al., Chauhan et al. , and other authors have already proposed probable blue-prints effective against the SARS-CoV -2. The authors have used different proteins and different epitopes in numerous combinations in their works and all these designed vaccines showed quite promising results in their experiments. However, in our work, novel polyvalent vaccines were designed that might be effective against the four most dangerous strains of HCoVs i.e., HCoV-OC43, HCoV-SARS, HCoV-MERS, and SARS-CoV-2, targeting their highly accessible spike proteins. As a result, such vaccines will be effective against all the selected viral strains simultaneously and no separate vaccine will be needed for each of the strains. Such works on polyvalent vaccines are not very common but our findings should definitely open new avenues to design epitope based polyvalent vaccines to fight against these lethal viruses.

Overall, this study recommends V3 vaccine as the best vaccine construct based on the strategies employed in the study to be an effective countermeasure against the four mentioned HCoVs. However, further research are suggested to finally determine and validate the immunogenicity, stability, safety, efficacy, and various physicochemical characteristics of the suggested vaccines of this study.

The Coronaviridae family of viruses is responsible for many outbreaks over the past two decades and the current pandemic caused by the SARS-CoV-2 has already taken the lives of millions of people. Scientists are working hard and soul to develop effective vaccines against all these strains, although none of them generated sound and satisfactory outcomes. In our study, effective polyvalent vaccines were designed by mining the spike glycoproteins of the four most prevalent strains of the Coronaviridae family i. e., HCoV-OC43, SARS-CoV, MERS-CoV, and SARS-CoV-2. If positive results are achieved in further research, then our designed vaccines might have the capability to fight off these deadly viruses.

Human Coronaviruses (HCoVs) are a group of lethal viruses that has caused many outbreaks in the recent decades and at present the SARS-CoV-2 is the new strain of HCoV and its outbreak has become a global pandemic. The scientific community around world is racing for a satisfactory prevention measure to control the transmission of coronaviruses, especially the spread of SARS-CoV-2. In this study, epitopebased polyvalent vaccines were developed that might confer immunogenic protection against the four strains of HCoV i.e., HCoV-OC43, HCoV-MERS, HCoV-SARS, and SARS-CoV-2. Since the vaccines contained multiple T-cell as well as B-cell epitopes from all of these four viruses, they are expected to provoke both humoral and cell-mediated immunogenic responses within the body. Results of different experiments that were conducted in the study indicated that these polyvalent vaccines should be quite safe, effective, and responsive to use. However, since all these predictions were done based on the computational methods, more wet lab-based research is needed to finally confirm the outcomes of this study. With high cost requirements and multiple limitations for developing the live, attenuated or inactivated vaccine preparation for contagious agents like the HCoV, these peptide-based vaccine candidates might be relatively cheap and effective options to reach the entire world to combat the HCoVs.

Coronavirus infections-more than just the common cold

Van der hoek L. Identification of new human coronaviruses

Bats and coronaviruses

Homology-based identification of a mutation in the coronavirus RNA-dependent RNA polymerase that confers resistance to multiple mutagens

The molecular biology of coronaviruses

Severity and outcome associated with human coronavirus OC43 infections among children

SARS and MERS: recent insights into emerging coronaviruses

Chest ct findings in coronavirus disease-19 (covid-19): relationship to duration of infection

SARS and other coronaviruses as causes of pneumonia

Coronavirus infections and immune responses

Role of the spike glycoprotein of human middle east respiratory syndrome coronavirus (mers-cov) in virus entry and syncytia formation

Identification of sialic acid-binding function for the Middle East respiratory syndrome coronavirus spike glycoprotein

Coronavirus spike proteins in viral entry and pathogenesis

Structure, function, and evolution of coronavirus spike proteins

Development of epitopebased peptide vaccine against novel coronavirus 2019 (SARS-COV-2): immunoinformatics approach

Identifying candidate subunit vaccines using an alignment-independent method based on principal amino acid properties

VaxiJen: a server for prediction of protective antigens, tumour antigens and subunit vaccines

T cell receptor sharing by cytotoxic T lymphocytes facilitates efficient virus control

Cd4 t cells: fates, functions, and faults. blood

The immune epitope database (IEDB): 2018 update

The early history of B cells

AllergenFP: allergenicity prediction by descriptor fingerprints

AllerTOP-a server for in silico prediction of allergens

Consortium, osdd; raghava, gps in silico approach for predicting toxicity of peptides and proteins

Identification of epitope-based peptide vaccine candidates against enterotoxigenic escherichia coli: a comparative genomics and immunoinformatics approach

Designing of interferon-gamma inducing MHC class-II binders

Computer-aided designing of immunosuppressive peptides based on IL-10 inducing potential

Evaluation of methods for the prediction of membrane spanning regions

Overlapping cd8+ and cd4+ t-cell epitopes identification for the progression of epitope-based peptide vaccine from nucleocapsid and glycoprotein of emerging rift valley fever virus using immunoinformatics approach

Predicting population coverage of T-cell epitope-based diagnostics and vaccines

PEPFOLD3: faster de novo structure prediction for linear peptides in solution and in complex

Improved PEP-FOLD approach for peptide and miniprotein structure prediction

PEP-FOLD: an updated de novo structure prediction server for both linear and disulfide bonded cyclic peptides

Dassault systèmes biovia, discovery studio visualizer, v. 17.2, san diego: dassault systèmes

A novel design of a multi-antigenic, multistage and multi-epitope vaccine against helicobacter pylori: an in silico approach

Recent advances of vaccine adjuvants for infectious diseases

Design of the linkers which effectively separate domains of a bifunctional fusion protein

Development of a multi-epitope peptide vaccine inducing robust T cell responses against brucellosis using immunoinformatics based approaches

Vaccination with a paramyosinbased multi-epitope vaccine elicits significant protective immunity against trichinella spiralis infection in mice

Discovery and investigation of O-xylosylation in engineered proteins containing a (GGGGS) n linker

Toll-like receptors, chemokine receptors and death receptor ligands responses in SARS coronavirus infected human monocyte derived dendritic cells

A multi-subunit based, thermodynamically stable model vaccine using combined immunoinformatics and protein structure based approach

Modulating immunity as a therapy for bacterial infections

Immune adjuvant effect of molecularly-defined toll-like receptor ligands. Vaccines (Basel)

Human β-defensin -3 activates professional antigen-presenting cells via toll-like receptors 1 and 2

A potential protein adjuvant derived from mycobacterium tuberculosis rv0652 enhances dendritic cells-based tumor immunotherapy

Immunoinformatics analysis and in silico designing of a novel multi-epitope peptide vaccine against staphylococcus aureus. infection

Differential expression of miRNA regulates T cell differentiation and plasticity during visceral leishmaniasis infection

Improving therapeutic HPV peptide-based vaccine potency by enhancing CD4+ T help and dendritic cell activation

High-throughput prediction of protein antigenicity using protein microarray data

S: algPred: prediction of allergenic proteins and mapping of ige epitopes

SOLpro: accurate sequence-based prediction of protein solubility

Protein-sol: a web tool for predicting protein solubility from sequence

The PSIPRED protein analysis workbench: 20 years on

Protein secondary structure prediction based on position-specific scoring matrices

GOR method for predicting protein secondary structure from amino acid sequence

Improving the accuracy of protein secondary structure prediction using structural alignment

Template-based protein structure modeling using the raptorx web server

Protein threading using context-specific alignment potential

RaptorX: exploiting structure information for protein alignment by statistical inference

Evaluation of predictions in the CASP10 model refinement category

Novel immunoinformatics approaches to design multi-epitope subunit vaccine for malaria by investigating anopheles salivary protein

GalaxyWEB server for protein structure prediction and refinement

Stereochemical quality of protein structure coordinates

ProSA-web: interactive web service for the recognition of errors in three-dimensional structures of proteins

Disulfide by design 2.0: a web-based tool for disulfide engineering in proteins

Amino acid neighbours and detailed conformational analysis of cysteines in proteins

HLA-DR: molecular insights and vaccine design

Toll-like receptors regulation of viral infection and disease

Toll-like receptors in antiviral innate immunity

Coronaviruses: an overview of their replication and pathogenesis. in coronaviruses

New additions to the c lus p ro server motivated by capri. proteins: structure

The cluspro web server for proteinprotein docking

How good is automated protein docking?

HawkDock: a web server to predict and analyze the protein-protein complex based on computational docking and MM/GBSA

Assessing the performance of the mm/ pbsa and mm/gbsa methods. 1. the accuracy of binding free energy calculations based on molecular dynamics simulations

HawkRank: a new scoring function for protein-protein docking based on weighted energy terms

Assessing the performance of mm/pbsa and mm/gbsa methods. 4. accuracies of mm/pbsa and mm/gbsa methodologies evaluated by various simulation protocols using pdbbind data set

ElliPro: a new structure-based tool for the prediction of antibody epitopes

Mutation-structure-function relationship based integrated strategy reveals the potential impact of deleterious missense mutations in autophagy related proteins on hepatocellular carcinoma (HCC): A comprehensive informatics approach

Monomerization alters the dynamics of the lid region in campylobacter jejuni cstii: an md simulation study

iMODS: internal coordinates normal mode analysis server

iMod: multipurpose normal mode analysis in internal coordinates

Predictions of protein flexibility: first-order measures

Computational immunology meets bioinformatics: the use of prediction tools for molecular binding in the simulation of the immune system

Exploring leishmania secretory proteins to design b and t cell multi-epitope subunit vaccine using immunoinformatics approach

Codon usage: nature's roadmap to expression and folding of proteins

JCat: a novel tool to adapt codon usage of a target gene to its potential expression host

Subtractive proteomics to identify novel drug targets and reverse vaccinology for the development of chimeric vaccine against acinetobacter baumannii

Application of a recombinant capsid polyprotein (P1) expressed in a prokaryotic system to detect antibodies against foot-and-mouth disease virus serotype O

Mfold web server for nucleic acid folding and hybridization prediction

Expanded sequence dependence of thermodynamic parameters improves prediction of RNA secondary structure

RNA secondary structure prediction

The vienna RNA websuite

Designing a multi-epitope based vaccine to combat kaposi sarcoma utilizing immunoinformatics approach

More than one reason to rethink the use of peptides in vaccine design

Immuno-informatics: mining genomes for vaccine components

Multi-epitope vaccines: a promising strategy against tumors and viral infections

Reverse vaccinology 2.0: human immunology instructs vaccine antigen design

Calculation of protein extinction coefficients from amino acid sequence data

How to measure and predict the molar absorption coefficient of a protein

Correlation between stability of a protein and its dipeptide composition: a novel approach for predicting in vivo stability of a protein from its primary sequence

Thermostability and aliphatic index of globular proteins

Analysis and prediction of highly effective antiviral peptides based on random forests

The role of antibody and complement in the control of viral infections

CD4+ regulatory T cells: mechanisms of induction and effector function

Introduction to T and B lymphocytes

Structural basis of T cell recognition

Physicochemical characterization and functional analysis of some snake venom toxin proteins and related non-toxin proteins of other chordates

Multi epitopes vaccine prediction against severe acute respiratory syndrome (sars) coronavirus using immunoinformatics approaches

IL-4-secreting CD4+ T cells are crucial to the development of CD8+ T-cell responses against malaria liver stages

In-silico design of a multi-epitope vaccine candidate against onchocerciasis and related filarial diseases

Metagenomic deep sequencing reveals association of microbiome signature with functional biases in bovine mastitis

Atypical MHC class II-expressing antigen-presenting cells: can anything replace a dendritic cell

Expression of sars-coronavirus nucleocapsid protein in escherichia coli and lactococcus lactis for serodiagnosis and mucosal vaccination

Synonymous codon usage pattern in glycoprotein gene of rabies virus

The importance of mRNA structure in determining the pathogenicity of synonymous and non-synonymous mutations in haemophilia

Design an efficient multi-epitope peptide vaccine candidate against sars-cov-2: an in silico analysis. Infect Drug Resist

Epitope based vaccine prediction for SARS-COV-2 by deploying immuno-informatics approach

Immunoinformatics study to search epitopes of spike glycoprotein from sars-cov-2 as potential vaccine

Excavating SARS-coronavirus 2 genome for epitope-based subunit vaccine synthesis using immunoinformatics approach

Authors are thankful to Swift Integrity Computational Lab, Dhaka, Bangladesh, a virtual platform of young researchers, for providing the support and tools.

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.