Discovery of potential immune epitopes and peptide vaccine design - a prophylactic strategy against Rift Valley fever virus

Proteome collection and antigenicity evaluation

The 232 complete proteomes of RVFV were obtained from ViPR (Virus Pathogen Resource) database (Supplementary Table 1, Extended data)²³. ViPR is a reliable and open database for pathogenic virus families²⁴. The viral proteome sequences retrieved were evaluated using VaxiJen v2.0 server for scores indicating antigenic influence. The threshold for antigenicity was fixed at 0.5²⁵. VaxiJen v2.0 sever possesses high prediction powers and utilizes auto cross-covariance (ACC) transformation methods. The glycoprotein sequence with the highest antigenicity score was taken from this proteome for further analysis.

CTL epitopes and their MHC class I alleles predictions

the NetCTL v1.2 tool was utilized to isolate CTL peptides from the RVF glycoprotein sequence. This server generates different nonamer epitopes against 12 supertypes (A1, A2, A3, A24, A26, B7, B8, B27, B39, B44, B58, B62). NetCTL v1.2 identifies CTL epitopes depending on C-terminal cleavage, TAP competence and MHC-1 complex binding²⁶. The threshold was determined at 0.5 with a corresponding sensitivity of 0.89 and specificity of 0.94. The MHC-I affinity for the CTL peptides was identified using the MHC-I IEDB web server²⁷ and a consensus percentile of ≤5 was set to narrow down CTL epitopes, for obtaining high-affinity epitopes with MHC-1 alleles.

Assessment of CTL epitopes

The isolated CTL epitopes were assessed for antigenicity by submitting them to the VaxiJen v2.0 server. To ensure proper induction of immune response in the human body, the immunogenicity of these epitopes was evaluated using the IEDB Immunogenicity tool²⁸. Furthermore, AllerTOP v2.0 server was chosen for predicting allergenicity and ensure that the vaccine construct does not induce an allergic reaction in humans. AllerTOP v.2.0 uses k-nearest neighbors (kNN) methods to distinguish allergens from non-allergens²⁹. Lastly, the selected nonamers were screened for probable toxicity using ToxinPred server, which predicts toxic epitopes using quantitative matrix methods and machine learning technology³⁰.

Identification of HTL epitopes and binding MHC class II alleles

HTL acts as the orchestrator to stimulate B cells, macrophages, and CD8+ cells to fight against pathogens. Consequently, HTL epitopes are important for making an effective vaccine³¹. HTL epitopes, each 15-mer in length, were identified by utilizing the IEDB MHC-II tool. The anticipation of binding of the epitopes to class II alleles, HLA-DP, HLA-DR and HLA-DQ, were determined utilizing the consensus 2.22 prediction method on the same server³². Due to obtaining a large number of epitopes, a percentile rank of ≤0.3 was set as a threshold.

Selection of cytokine-inducing HTL epitopes

HTLs release cytokines such as interferon-gamma (IFN-γ), interleukin 4 (IL-4), and interleukin 10 (IL-10) that can activate immune cells in the body. In addition, cytokines released by HTLs can survive past inflammatory responses and avert tissue damage. Therefore, HTL epitopes that can induce the release of cytokines are important in vaccine development. Therefore, in order to incorporate the epitopes that induce IFN- γ, we used the IFNepitope server, through the hybrid method (motif and SVM)³³. Furthermore, the peptides were evaluated to check if they induce IL-4 and IL-10 using IL4pred and IL10pred servers, respectively^34,35.

Evaluation of LBL epitopes

The activation of B cells plays a vital part in the activation of the humoral immune response and generation of plasma cells against a specific antigen, and distinguishing LBL epitopes is another vital step in the development of epitope-based vaccine constructs. LBL epitopes were predicted using a combinatorial algorithm of gradient boosting and a randomized tree method using the iBCE-EL server³⁶. The predicted LBL epitopes were then reevaluated for their antigenicity score with VaxiJen v2.0 server, and non-allergenic and non-toxic epitopes were predicted using the AllerTOP v2.0 server and ToxinPred tool, respectively.

Estimation of population coverage

For an epitope to be antigenic and evoke a strong immune reaction it needs to be acknowledged by the MHC complex molecule. MHC alleles are the most polymorphic and occur in thousands of HLA combinations in humans. Therefore, an HLA allele with a high frequency of occurrences in the majority population of the world would have a high chance of exerting an effect immunogenic³⁷. In our study, we wanted to find out the distribution, presence and frequency of the T cell epitopes, which were selected for the purpose of designing the vaccine structure. Allele Population Coverage of IEDB population coverage tool was utilized for the calculation of population coverage³⁸. A prediction analysis was run on regions of Africa where the RVFV outbreak initially started and the neighboring countries that were mostly affected and also across the entire world.

Assembling of multiepitope vaccine

All the predicted and assessed epitopes i.e. CTL, LBL, and HTL peptides were joined together using linkers and an adjuvant sequence was added upstream of all of them to form a vaccine structure. 50S ribosomal protein L7/L12 (GenPept Accession: P9WHE3) is a toll-like receptor 4 (TLR4) agonist and was chosen as an adjuvant for the construct to boost the immune response against it²¹. TLR4 can recognize viral glycoproteins and bacterial ligands^39,40 stimulates the production of cytokines against them⁴¹. The CTL peptides and the 50S L7/L12 protein adjuvant were linked together with the EAAAK linker to ensure adequate separation between each component for their effective interaction with their respective targets. This linker was chosen as it separates bifunctional fused protein domains⁴². Each CTL epitope was separated using AAY linkers, and HTL epitopes by GPGPG linkers. Independent immunogenic potential of each LBL epitope was preserved by separating them using KK linkers⁴³.

Host homology crosschecking

To avoid the risk of elicitation of autoimmunity by molecular mimicry, we screened the vaccine construct against the Homo sapiens protein sequences (NCBI: txid9606) through the use of NCBI BLASTp⁴⁴.

Physicochemical evaluation of the vaccine construct

Vaccines are developed to provide protection against diseases by evoking an immune response after injection against specific antigens, without causing any disease. Thus, vaccines need to have antigen inducing capability without being allergenic to the receiver. Therefore, the multiepitope vaccine construct was tested for antigenicity and allergenicity in each step of the initial design, using VaxiJen v2.0²⁵ and AllerTOP v2.0 servers, respectively²⁹. Predictive assessment of different physiochemical attributes of the subunit vaccine was carried out using the ProtParam server⁴⁵. Lastly, the SOLpro server in the SCRATCH suite was utilized to assess the solubility of the vaccine structure in E. coli, with a view to determining the bioavailability of the vaccine⁴⁶.

Secondary and tertiary modelling of the construct

Predicted two dimensional (2D) configuration of our vaccine design was generated using the PSIPRED v4.0 web tool. The PSIPRED tool utilizes the query amino acid residues to predict a 2D model using two-feed-forward neural networking along with PSI-BLAST⁴⁷. The three dimensional (3D) model of our designed vaccine was attained through the I-TASSER web tool, which exhibits protein modeling via a hierarchical procedure to estimate and bring about suitable structure and function⁴⁸. The I-TASSER site helps to generate 3D structure of a protein and determine its functions using a state-of-the-art algorithm with high precision. This web tool enables determination of C-score, TM-score value and root mean square deviation (RMSD), along with the top five predicted structure models of the given protein sequence⁴⁸. The produced 3D structure was chosen based on its C-score value and downloaded in PDB format. The server provides a C-score ranging between -5 to 2, where a higher value indicates a better model⁴⁸.

Refinement and verification of 3D vaccine

Our 3D vaccine structure was refined using the GalaxyRefine server. The web-based program reconstructs sidechains, then repacks them using dynamic simulations for proper structural relaxation⁴⁹. The ProSA tool was used to recognize possible errors in the predicted tertiary model and for structural validation¹⁶. If the calculated score of ProSA deviates from the given range, this indicates that the primary sequence and predicted structure possibly contain flaws⁵⁰. Additionally, the Verify3D server was employed to examine the congruity of a tertiary structure with its primary sequence. It is done by specifying a basic grade depending upon its position and surroundings and correlating the findings to known verified structures⁵¹. Ramachandran plot generation was carried out using the RAMPAGE tool⁵². The Ramachandran analysis plot is a visual representation of energetically permitted and rejected dihedral planar angles based on weak Van der Walls force of interactions between amino acids of the side chain. The RAMPAGE score enumerates the residues residing in favored, allowed and disallowed regions (in %) based on the PROCHECK principle⁵³.

Disulfide bridging of vaccine

Disulfide bonds are intra-protein bonds that can help stabilize tertiary or quaternary interactions in a protein⁵⁴. Therefore, disulfide engineering was accomplished by introducing two novel bonds in the vaccine framework by employing the Disulfide by Design v2.12 web tool. The refined vaccine complex was uploaded, checked for a residue-pair match within parameters of χ3 angle -87° or +97° ± 30 and Cα-Cβ-Sγ angle 114.6° ± 10. Appropriate amino acid duos were chosen and mutated to cysteine residues⁵⁵.

Codon optimization and in silico cloning

Codon refinement is a vital step for the expression of the protein inside the host system. Therefore, codon adaptation was carried out by utilizing the JCat web tool in order to improve protein translation. E.coli K12 was selected for cloning since unadjusted codon sequences will reduce protein expression⁵⁶. Termination of transcription (rho-independent), prokaryotic ribosome region of attachment and restriction enzyme cut-off sites were disregarded by setting the parameters to “avoid”. The server further provided two additional measures such as the codon adaptation index (CAI) and GC content of the refined sequences⁵⁷. The codon nucleotide adapted sequence was cloned using SnapGene v5.0.8 software into a E. coli pET28a(+) vector obtained from SnapGene⁵⁸. ApE and Genome Compiler are open access alternative to this software that could be used for this purpose.

Molecular docking between TLR4 and vaccine

Molecular docking is a computerized process that evaluates and assesses the contact between a protein and its receptor. The binding affinity and interaction between the proteins provide a simulated score⁵⁹. TLR4 can recognize viral glycoprotein⁶⁰ and therefore, the TLR4 structure was obtained from PDB (PDB ID: 4G8A) and was chosen as a receptor to dock with the refined vaccine⁶¹. ClusPro v2.0 web docking server was appointed to determine the binding position and inclination for binding between the designed vaccine and TLR4 receptor⁶². The vaccine-receptor composite was selected based on the binding position (on active site), docking efficiency and the lowest energy scoring of the docked complexes.

Molecular dynamics simulation

Determining the stability and firmness of the protein-receptor docked structure by molecular dynamics is essential for carrying out an in silico study. Protein stability was resolved by contrasting protein dynamics and their counterpart normal modes^63,64. the iMODS tool was used to evaluate and plot amino acid residues and their motion within their inner order via normal mode analysis (NMA)⁶⁵. The NMA tool surfaced the extent of motion of the vaccine-receptor complex in terms of covariance and ability to deform; along with eigenvalues and B-factor results. The deformability of the complex depends on whether it can rotate each of its residues while staying in its position. The eigenvalue demonstrates the rigidity of motion, being directly proportional to the energy required for the deformation of the structure. Lower eigenvalues represent easier structural deformation^65,66.

Estimation of immune response

Simulations of immune response of the vaccine design were executed with the C-IMMSIM v10.1 server. This server uses position-specific scoring matrix (PSSM) and AI-based technology to predict the intensity of immune reaction caused by the multiepitope vaccine⁶⁷. It evaluates the immunologic response of the subunit vaccine according to an in vivo system and can also simulate immune response through its agent-based dynamic system. The minimum recommended interval is at least four weeks between the first and second dose⁶⁸. Time-steps for each vaccine dose were fixed: 1, 84 and finally 168 respectively, with an interval of eight hours in-between. Analysis of three doses was conducted in respect to immune cell rise and activity with at least four weeks in between each of the dose sessions.

RVFV proteome and selected antigen

Amino acid sequences of a total of 232 RVFV proteins were collected from the ViPR database. The VaxiJen v2.0 server revealed the two proteins with the highest antigenicity (Table 1). The glycoprotein sequence was selected for further analysis.

Potential CTL, HTL and LBL epitopes

NetCTL v1.2 server predicted 135 unique 9-mer CTL epitopes in total. Out of them, 27 epitopes were predicted as having the ability to act as both antigen and immunogen and were also nontoxic and nonallergenic (Supplementary Table 2, Extended data)²³. Only six CTL peptides were shortlisted for the ultimate vaccine design based on a combined score (Table 2).

Similarly, 182 15-mer HTL epitopes and their MHC-II binding alleles were predicted and evaluated by utilizing the IEDB MHC-II tool. Next, cytokine production capability of the selected HTL epitopes was predicted for IL-10, IL-4 and IFN-γ. Only 14 epitopes were found to be capable of inducing any of the three cytokines’ production and to have antigenicity (Supplementary Table 3, Extended data)²³. Of those that were capable of producing IFN-γ, two epitopes were chosen with the ability to produce IL-4 but not IL-10 and one epitope was chosen with the ability to produce IL-10 but not IL-4. Therefore, three epitopes were finally shortlisted prioritizing for IFN-γ and IL-4 production (Table 3).

B cells act as antigen-presenting cells that recognize epitopes present on a protein and can bring about a humoral immune response⁶⁹. The iBCE-EL sever program was used to predict unique LBL epitopes from the RVFV glycoprotein sequence. A total of 310 LBL epitopes were found. After further evaluation, 35 unique epitopes were found to be non-allergenic and non-toxic (Supplementary Table 4, Extended data)²³. Amidst those, only five LBL epitopes were chosen for vaccine construction based on iBCE-EL server’s probability scores. LBL epitopes with higher probability scores were chosen. (Table 4).

Estimated population coverage

The CTL and HTL epitopes across African regions and Southwest Asia have a mean population coverage of 77.39 %, with an average hit of 1.78. The total combined class 1 and 2 epitopes have significant coverage across Europe (99.92%), North America (100%), South America (99.39%) and South Asia (99%), which indicates the multiepitope vaccine would be a good candidate in eliciting an immune response in these worldwide regions. In the African population, where the initial outbreak began, there is good coverage in the majority of regions, except for South Africa (18.36%), as shown in Figure 2.

Figure 2. Population coverage with selected epitopes and their respective alleles.

Epitope based subunit vaccine construct

To design our multiepitope vaccine, we picked four CTL epitopes (Table 2) based on their high immunogenicity and antigenicity scores and which were non-allergic and nontoxic. For the HTL epitope selection, we looked for epitopes that could produce all three types of cytokine; however, no epitope was found that had the capacity to induce all three types of cytokines (IFN-γ, IL-10, and IL-4). Hence, only two HTL epitopes were chosen from the glycoprotein sequence, which were capable of inducing IFN-γ and IL-4 (Table 3). Three LBL epitopes that were nontoxic and non-allergic and had the best iBCE-EL predicted probability, high antigenicity scores were selected (Table 4). Two extra epitopes were shortlisted for each of the three cases for the purpose of randomization to find the optimum vaccine construct. The choice of adjuvant was 50S ribosomal protein L7/L12, retrieved from the NCBI Protein database (Accession no. P9WHE3) and was linked to the selected epitopes using an EAAAK linker. For the final vaccine construct, four out of six CTL, two out of four HTL and three out of five LBL shortlisted epitopes were chosen based on different combinations and randomization to generate seven potential vaccine candidates, provided in Supplementary Table 5 (see Extended data)²³, and were merged with AAY, GPGPG and KK linkers, respectively (Figure 3). The prospective vaccine with the ideal physicochemical property was chosen. The final assembled vaccine is 262 amino acid residues long.

Figure 3. Diagram of the assembled vaccine construct outlining 262 amino acids.

The EAAAK linker was utilized to join the adjuvant at the 5’ end to the rest of the vaccine. CTL epitopes linked with AAY linkers, HTL epitopes linked with GPGPG linkers and form a bridge with the last CTL epitope and first HTL epitope. KK linkers were used to connect LBL epitopes and also form a bridge between the last HTL epitope and the first LBL epitope.

LBL, linear B-lymphocyte; HTL, helper T-lymphocyte; CTL, cytotoxic T-lymphocyte.

Host-pathogen antigenic similarity

Evaluation of the multiepitope vaccine protein for similarity against the Homo sapiens proteome was done using BLASTp to ensure safety. No similar proteins were found, which negates the chance of autoimmunity as a result of the vaccine.

Physicochemical properties and solubility

Physiochemical attributes of the vaccine structure were assessed using the ProtParam tool. The vaccine’s molecular weight was estimated to be ~28 kD and showed good antigenic characteristics. The predicted Isoelectric point (pI) of the designed vaccine is 4.96, indicating an acidic nature, and it has an instability index of 27.79, suggesting that the vaccine will maintain good stability inside the host system. The aliphatic index is 84.73, denoting the thermostable nature of the vaccine. The hydrophilic nature of the vaccine was estimated by the grand average of hydropathicity (GRAVY) score of -0.125. Half-life (t_1/2) was calculated to be greater than 30 hours in a mammalian reticulocyte, greater than 20 hours in yeast, and more than 10 hours in E. coli. The overall vaccine construct is not allergenic and demonstrated good solubility as shown by the SOLpro server. The above-mentioned assessments suggest that our multiepitope vaccine design has the potential to be a good candidate for RVFV (Table 5).

Secondary structural features

The 2D structure of the vaccine design was acquired through the PSIPRED v4.0 workbench. The secondary structure consisted of 49.62% α-helices (130 amino acids), 11.83% β-strands (31 amino acids) and 38.55% random coils (101 amino acids) (Figure 4).

Figure 4. Secondary structural features of the designed vaccine.

Herein, α-helix, β-strands, and random coils are represented with pink, yellow and blue colors, respectively.

Refined 3D structure of the vaccine

The multiepitope vaccine was speculated and constructed using I-TASSER server. The PDB structure (PDB ID: 1DD4) was regarded as the best template for modeling. This server carries out 3D modeling based on the consequence of threading template alignment and simulations are run by merging parameters of the structure assembly. It ranks the confidence of models quantitatively on C-score. The obtained C-score for the modeled 3D vaccine structure was -4.16. The refinement of the 3D structure was carried out by the GalaxyRefine web-tool. The refined and polished model was used to generate a Ramachandran plot, which showed an increased percentage of residues in the favored region. The refined vaccine construct showed 90.0% residues in the favorable region in Ramachandran plot. The global distance test (GDT-HA) score was found to be 0.9055, RMSD was 0.534, MolProbity was 2.209, clash score was 14.2 and poor rotamers were non-existent (Figure 5).

Figure 5. The tertiary structure of the vaccine where α-helix, β-strand, and random coils are shown in red, yellow, and green, respectively.

Validation properties of the 3D vaccine

The refined tertiary structure was validated using the RAMPAGE server, ProSA-Web tool and Verify3D server. Validation carried out by the RAMPAGE server using a Ramachandran plot of the unrefined 3D structure showed 69.2% of residues in the favorable region, 20.8% in acceptable regions, and 10.0% of the residues in disallowed regions. After structural refinement was carried out, RAMPAGE revealed 90.4% of the vaccine’s residues resided in favorable regions, and 6.2% and 3.5% of residues were found in acceptable and disallowed regions, respectively (Figure 6A). The ProSA-web and Verify3D (Figure 7) servers were used to validate the unrefined tertiary structure. After refinement, ProSA-web revealed a Z-score of -6.2 for the best vaccine protein model, implying equivalence to the native protein conformation (Figure 6B).

Figure 6. Qualification of our predicted 3D arrangement of the vaccine.

(A) Ramachandran plot analysis of the refined model showing favored, allowed and disallowed regions are 90.4%, 6.2% and 3.5%, respectively. (B) Validation using ProSA web tool, revealing a Z-score of -6.2.

Figure 7. Verify3D shows a score of 80.92% for our selected vaccine model.

Disulfide bridging in the vaccine structure

Disulfide bridging was used to stabilize the refined vaccine model. The introduction of cysteine residues was carried out via Disulfide by Design v2.12. Although 23 potential pairs of residues were found (Supplementary Table 7, Extended data)²³ suitable for disulfide engineering, only two pairs of residues were chosen based on the energy of binding and χ3 angle. Accordingly, two mutations were created on a pair of selected amino acids, based on the lowest kcal/mol value. For, Leu67-Ala100 residues, the energy value was 2.40 kcal/mol and the χ3 angle was -103.13 degrees and for Phe28-Ala37 duos, the χ3 angle was -111.70 degrees and energy value was 2.86 kcal/mol (Figure 8).

Figure 8. Disulfide bridging of the vaccine model to increase stability.

The engineered pairs are indicated by blue and red colors.

Adapted codons and in silico cloning

The primary objective is to express our constructed multiepitope vaccine sequence using in silico cloning into the E. coli expression system. As such, the subunit vaccine construct requires codon optimization according to the codon usage for the expression pattern system in E. coli. To optimize for maximal expression of our vaccine design in E. coli K12, the JCat tool was used. The optimized codon sequence was 786 base pairs in length. The GC content was 49.75%, which lies within the ideal range of 30–70%. Codon Adaptation Index was 1.0, which is also within the ideal range (0.8–1.0). This ensures the efficient expression of the multiepitope vaccine inside E. coli. XhoI and BamHI restriction sites were inserted into E. coli plasmid pET28a(+) and the adapted vaccine’s codon sequences were inserted into a recombinant E. coli pET28a(+) plasmid using SnapGene software v5.0.8 (Figure 9). The final length of the recombinant construct is 6121 bp.

Figure 9. Recombinant plasmid construct pET28a(+) containing multiepitope vaccine was inserted between XhoI (158 bp) and BamHI (950 bp) restriction sites shown in red.

Outer dark margin represents the vector skeleton.

Molecular docking studies

The ClusPro v2.0 server tool was used to conduct molecular docking for characterizing contact between the illustrated refined vaccine and TLR4 (PDB ID: 4G8A). This server provided 30 different docked models, the best 10 of which were analyzed for the selection of the appropriate vaccine-receptor complex (Supplementary Table 6, Extended data)²³. The model showing a strong interaction between vaccine residues and the active site of TLR4 along with the lowest energy value was selected. Model 7 fulfilled the above-mentioned criteria, and thus, was picked as the suitable vaccine-receptor complex (Figure 10). Model 7 had an energy score of -809.1 and indicated the highest binding affinity as it had the lowest energy score (Supplementary Table 6, Extended data)²³.

Figure 10. Interaction between the vaccine and TLR4 receptor.

Vaccine surface is represented in yellow, while the receptor is represented in cyan.

Molecular dynamics simulation

To understand the vaccine-receptor compound stability and its large-scale mobility, NMA was carried out using iMODS tool, which is simulated based on intramural configuration of the whole compound. Deformability of the vaccine-immune receptor complex depends on the isolated movement of every amino acid residue, illustrated via chain hinges (Figure 11). The eigenvalue was 2.443455e-5⁵⁷ (Figure 11A). The covariance matrix demonstrates the linkage between duplets of amino acids, the correlated residues marked in red, anti-related residues in white and non-correlated residues in blue, interspersed in dynamical zones (Figure 11B). The elastic mesh-work paradigm classifies which pairs of atoms are interlinked by springs and is visualized as a linking matrix. Dots are colored in accordance to their rigidity; the higher the rigidity, the darker the color (Figure 11C)⁷⁰.

Figure 11.

Molecular dynamic simulation of vaccine-receptor complex, showing (A) eigenvalue; (B) covariance matrix; and (C) elastic network analysis.

In silico immune responses

The simulated immune response showed resemblance to an actual immune response against a pathogen (Figure 12). Subsequent exposures produced a higher tier immune response compared to the primary immune response. Secondary and tertiary responses showed higher levels of antibodies (i.e., IgM, IgG1, IgG2), which coincides with the waning of antigen (Figure 12A). This demonstrates the development of immune memory cells and, as a result, intensified antigen neutralization upon successive exposure (Figure 12A). Additionally, several B-lymphocyte cell isotypes with prolonged life have been noticed, signifying potential class switching and memory B cell generation (Figure 12B–12C). Increased responses were also observed in helper T cells and cytotoxic T cells with memory generation (Figure 12D–12F). An increase in macrophage activity and engagement was perceived, with a vigorous proliferation of dendritic cells (Figure 12G–12H). An increased amount of IFN-γ and IL-2 cytokines were also noticed (Figure 12l). These findings suggest the development of immune memory and, therefore, immunity against the virus.

Figure 12.

In-silico immune response for vaccine construction as an antigen using C-IMMSIM server tool: (A) production of immunoglobulin due to injection of the vaccine; (B) B cell count following consecutive three injections; (C) population of B cell number per state; (D) helper T cell activation; (E) population of helper T cell number per state; (F) population of cytotoxic T cell number per state; (G) per state macrophage population; (H) dendritic cell population; and (I) level of cytokine and interleukins.

Underlying data

All data underlying the results are available as part of the article and no additional source data are required.

Extended data

Harvard Dataverse: Extended DATA: Discovery of potential immune epitopes and peptide vaccine design - a prophylactic strategy against Rift Valley Fever Virus. https://doi.org/10.7910/DVN/B9Y1EH²³.

This project contains the following extended data in DOCX format:

Data are available under the terms of the Creative Commons Zero “No rights reserved” data waiver (CC0 1.0 Public domain dedication).

Author endorsement

Mohammad Minnatul Karim confirms that the author has an appropriate level of expertise to conduct this research, and confirms that the submission is of an acceptable scientific standard. Mohammad Minnatul Karim declares they have no competing interests. Affiliation: Department of Biotechnology and Genetic Engineering, Islamic University, Bangladesh.