Principal Component Analysis. PCA was the final analysis to visualize DENV envelope protein variation patterns. The biplot in Figure 6 describes the variability in PC1 and PC2. The PCA obtained four distinct and well-separated groups, and each group represented one of the serotypes of the virus (DENV-Type I to IV). The variants explained by PC1 was 36.08%, while PC2 was 27.19%, and the total captured by the first two components, PC1 + PC2 = 63.27%, explains the variation in envelope protein sequences, which is why the plot clearly separates the clusters. The isolates of DENV Type 1 (green circles in the upper left corner) have the highest coordinates in PC2, and the EMRO strains from 2023 to 2025 cluster together with the strains from India and China. Though the DENV-2 isolates (right cluster, brown triangle) were clumped together with less PC1 variation, hence low variability within DENV-2 isolates despite being geographically diverse, among them, more recent isolates such as UAE 2024, Pakistan 2025, and Somalia 2023, overlapped with other recent isolates, including India 2025 and Cuba 2024, which supported the heatmap findings. For the DENV-3 isolates (bottom-left quadrant, blue squares), they clumped together, with this cluster encompassing both old and new isolates. Interestingly, the Cuban GIIIAII strain (PQ508793.1, 2024) and some recent strains isolated in Israel and Mali were located towards the outskirts of the group. Likewise, the DENV-4 strains (pink cross-markings) were clustered together, relatively closely and compactly, yet separate from other types. Overall, the results of the PCA are consistent with previous findings from the heat map and phylogenetic analysis: recent strains, particularly those from EMRO countries, are geographically and temporally clumped.

FIGURE TEMPLATE

3. Discussion

Comprehensive genomic data from the region remains limited and uneven, although the dengue virus has circulated in parts of the Middle East for over two decades. Studies like Hill et al. (2025) and Al-Eitan (2024) highlighted the challenges of monitoring mosquito-borne diseases (MBDs) in the region. Regional research, according to Hill et al., tends to be data-based upon sporadic outbreaks and a phylogenetic structure, thereby being unable to trace a distinct pattern for Arthropod-borne viruses (17, 18). In this analysis, it was found that Pakistan has the most up-to-date DENV sequence in the EMRO area, whereas the remaining countries have underrepresented sequences in public databases despite having confirmed cases, including Iran, Egypt, and Jordan—the issue poses a challenge for monitoring mutations within the virus and emerging strains in the region. First, we found that the DENV-2 sub-clade represents the greatest diversity in this area with newly found strains from Pakistan in 2023-2025 grouping together with strains from Somalia, UAE, and Djibouti aligning with existing data in the field, for example, genetic structure and distribution in Kenya of dengue virus-2, proving the exchange of genes between it and other countries through gene flow in East Africa (19). Research on Aedes aegypti mosquitoes in the Arabian Peninsula has shown that populations in Saudi Arabia originated from colonization by both the ancestral African form (A. aegypti formosus) and the global domestic form (A. aegypti aegypti), suggesting a viable transportation route for the dengue vector from Africa to the Arabian Peninsula (20). A 2024 study by Mashlawi et al. supported this by providing further genetic evidence of this dual origin, underscoring significant admixture between the two forms across different Saudi regions (21). Finally, a previous systematic review analysis from 2016 by Humphrey et al. discussed numerous reports that came from the Red Sea subregion and Pakistan, scoring the highest seroprevalence estimates for dengue, which indicated the significant virus circulation and potential for cross-border transmission in these areas (22). This research, together with our analysis results, supports the concept that East Africa and the Arabian Peninsula are intercontinental hubs for dengue movement. On a broader scale, integration of recent global sequences revealed no entirely new clades but rather a subtle shift in regional connectivity; note in the figure how the EMRO strains increasingly overlap with newer sequences from India, Cuba, and China 2024-2025, particularly in DENV-2 and DENV-3-meaning towards a possible recent transmission patterns and ongoing regional spread. Some India and China articles in 2025 support our observation of increased regional connectivity in dengue virus transmission, particularly involving DENV-2 and DENV-3 serotypes; for instance, India accounts for a substantial portion of the global dengue burden as the virus load is escalating due to the climate change; while China reported that dengue cases were found to be expanding and spreading across new area with imported cases primarily originating from Southeast Asian countries (23, 24). The inclusion of the Cuban DENV-3 sequences was classified under the emerging GIIIAII lineage in the Guzman et al. (2025) study, this new lineage, which was first detected in Cuba amid a spike in cases and supported by national molecular surveillance data, was shown to cluster closely with sequences from the USA, Puerto Rico, and notably, India (11). More importantly, the Cuban 2022 sequences exhibited close phylogenetic proximity to EMRO strains when placed in our broader phylogenetic framework, particularly the 2024 Israeli and 2021 Saudi Arabian isolates, indicating a possible shared transmission network or regional dissemination pattern not previously documented. In Saudi Arabia, a cross-sectional study by Melebari et. al. conducted between April 2023 and May 2024 in Makkah hospitals confirmed 238 dengue cases and found 11 unclassified samples that didn't fit into any dengue category, which highly supports the possibility of having novel or divergent strains (1). As of June 2025, there have been no reports of new dengue outbreaks or surveillance findings in Israel, aside from their previously reported imported cases, although some sequences have still been submitted. In addition, their most recent information from Israel came in October 2023, when their Ministry of Health reported 30 dengue cases, all of which had traveled to other countries (25). The danger of possible local transmission in such occurrences is underscored, especially knowing that the presence of competent vectors in the region increases the danger of regional transmission to other neighboring nations. In terms of protein analysis, the amino acid tree and analysis identified sequences with pivotal substitutions not present in their previous sequences (Figures 4, 5, and 6). As we mentioned previously, E protein sequence variations can affect the virus's fitness, tropism, and immune evasion (26-28). However, the present study did not experimentally validate the functional effects of the identified amino acid substitutions, and the findings should be interpreted as hypothesis-generating observations that require confirmation through functional and structural studies. From our analysis, we find significant patterns of amino acid substitution across the whole E protein sequences of recent DENV isolates, pointing to a recent rapid phase of molecular evolution that may have been driven by immune selection pressures or vector/host factors. Notably, these patterns are similar to or closely resemble those within the 2025 India/China E protein sequences, which may suggest convergent evolutionary processes or may reflect transmission association. On a more universal scale, it is clear through these findings, as well as those preceding and recent, that the role of antigenic drift, particularly for the E protein, persists as a non-neutralizing process, as shown both in vitro and vaccine models independently (29-32). Given this, our results highlight the need for continued surveillance of E protein variants and functional studies to assess the phenotypic consequences of these substitutions. Such evidence is provided by PCA, in which serotyping is more significant, whereas after 2023, these data show an association between the E protein in the Asia Pacific and those in EMRO, particularly for DENV-2 and DENV-3. Genetic drift is the second major cause of these mutations, and it determines the evolution of DENV; this selection pressure could continuously flip the shape of DENV sequences, leading to sudden outbreaks or the silent transmission of the disease (33). The emergent pattern analyzed may raise concern if a new form is silently spreading across regions due to travel, changes in vector populations driven by climate change, or reduced genomic surveillance. However, given the current data limitations, especially the lack of sequences from critical EMRO countries post-2022, we cannot yet confirm whether this represents a new regional lineage or coincidental parallel evolution; therefore, continued genomic surveillance, supported by broader sequence sharing and regional coordination, will be essential to clarify these trends and detect true emergent lineages before they result in major outbreaks.

4. Biosafety and Biorisk Management Recommendations

The lack of antivirals and effective medications against the dengue virus makes it a considerable biological threat; as it spreads worldwide, well-thought-out strategies must be implemented to lessen the danger (34, 35). Vector control remains a viable process to decrease the spread of the dengue virus, done by the use of insecticides such as larvicide and adulticide, insect repellent, and/or chemical agents, like picaridin and N, N-Diethyl-meta-toluamide, to keep waters sterile (34). Only Denvaxia has been granted licensing as a vaccine for dengue, yet it is still not recommended for use on patients with no prior dengue virus exposure; several initiatives are also being undertaken to develop an effective vaccine against dengue using techniques such as live-attenuated chimeric recombinant virus, live-attenuated virus, inactivated virus, recombinant protein, and mRNA vaccines. The National Institute of Allergy and Infectious Diseases’ TV003/TV005 vaccine is currently going through the clinical trials phase (34, 36). The World Health Organization (WHO) created a multidisciplinary plan consisting of 5 core components divided into 11 pillars dubbed the 5Cs, focusing on emergency coordination to prevent future spreads and develop a robust way to mitigate risk; it also aims to create a collaboration in surveillance to control its spread. Further ensures that the laboratories have access to rapid and reliable tests for suspected dengue virus patients; it also includes community protection and access to safe and scalable care and countermeasures (37). WHO also reported that the virus spread to Africa by travelers from Southeast Asia and Latin America; this led to the spread of the disease in highly dense urban areas which points out the need for more efficient travel surveillance and vector control (e.g., screening or educating travelers) (38).

5. Study Limitations

While our paper showed how the dengue virus’s circulate in the EMRO region, there seems to be a notable lack of complete/partial genome sequences in many of its countries including Afghanistan, Bahrain, Egypt, Iran (Islamic Republic of Iran), Iraq, Israel, Jordan, Kuwait, Lebanon, Libya, Morocco, Oman, Qatar, Somalia, Sudan, Syrian Arab Republic, Tunisia, Yemen, and the Occupied Palestinian Territory; even though the WHO reported cases of dengue virus related infections in these countries. Despite this challenge, we tried to include a diverse list of countries from the EMRO region with several countries from other areas and continents for better resemblance, still a majority of our sequences came from Pakistan which could in turn, create a geographic bias that may result in the overrepresentation of the clusters found in Pakistan at the expense of other countries found in the region. Furthermore, a few recent sequences came from the Eastern Mediterranean, as only one sequence was found from 2025. In the protein-level study, only a handful of data points were represented in the viral envelope protein. Therefore, we focused on the envelope (E) protein because of its well-known roles in viral entry, immune system recognition of the virus, and its relevance to vaccines. Although the non-structural proteins NS1, NS2A, and NS4B are recognized for their importance in viral replication and immune modulation, they were excluded from the study to maintain the manuscript's focus; including them would have substantially extended the scope and complicated the analysis, potentially diluting the primary objective of tracking transmission dynamics through envelope protein variation. Although this study presents new information on the variation of dengue virus types in their sequences, further study using more complete genomic information from across the EMRO region could provide insights into dengue virus circulation within and beyond the region, which could aid in the prevention of outbreaks in the future.

6. Future Direction

The next steps should focus on expanding the dengue virus genomic database, especially in EMRO and surrounding regions that were underrepresented in the past. This would highlight the spread of the dengue virus globally and on its evolution over time, and its possible by developing a "centralized digital platform” like VectorSurv system-like centralized digital platform (37), where all EMRO countries can upload their sequence data, and this platform will automatically monitor mutation trends and generate dynamic risk assessments for vector-borne disease outbreaks. Functional analyses would also be required to evaluate the potential biological consequences of the most significant amino acid changes, particularly those occurring at conserved or immunologically important positions. Such research might include structural modeling studies, epitope analyses, and neutralization tests. Finally, a coordinated surveillance effort with standardized sequencing techniques would allow for real-time surveillance for new variants, which could help prevent dangerous outbreaks in the future.

7. Conclusion

This work offers a detailed and informative overview of the dengue virus (DENV) evolution in the Eastern Mediterranean Region (EMRO) and worldwide. Through the combination of genomics and protein-based analysis, we show that the latest DENV strains (2023-2025) in regions like Pakistan, the UAE, Djibouti, and Cuba display a considerable degree of genetic diversity while being closely related to the latest global strains. Phylogenetic trees based on the analysis of either the genetic and/or protein sequences clearly display the close grouping of the EMRO strains, in particular for DENV-2 and DENV-3; these patterns suggest regional and cross-continental transmission dynamics that are not always visible through traditional surveillance. Other methods, such as PCA clustering and time-dated Phylogenetic trees, further supported that the newer strains represented the cutting edge of recent evolution in the case of the recent evolution of the virus. The entry of the Cuban GIIIAII strain and the identification of other unclassified serotypes in other countries clearly point to the incompleteness of the classification system and the need to have real-time genomic data, especially when the possibility of its undetected evolution increases due to high temperatures and vector populations. All together, these findings emphasize that without robust genomic data and coordinated surveillance across borders, the region may fail to detect or contain emerging variants early enough to intervene; however, given the current data limitations, especially the lack of sequences from critical EMRO countries, we cannot yet confirm whether this represents a new regional lineage or coincidental parallel evolution. Therefore, future work should prioritize standardized sequencing, regional data platforms and functional studies of key mutations to assess their impact on antigenicity and outbreak potential, allowing real-time tracking of emerging variants, and help mitigate the risk of dangerous future outbreaks.

8. Methods

8.1 Inclusion and Exclusion Criteria

To investigate dengue virus diversity and evolutionary trends in the Eastern Mediterranean Region, we accessed sequences from the NCBI Virus database, which is known for its high-quality and trustworthy genomic data (14). The following keywords were used to specify the dengue virus and its types: "Dengue virus", "Orthoflavivirus denguei", "Dengue virus type 1", "Dengue virus type 2", "Dengue virus type 3", "Dengue virus type 4", "Dengue virus 4", and "Dengue virus group", we obtained a total of 52,675 sequences; therefore, a set of filtration methods were followed to refine the dataset. First, we identified the EMRO region according to the World Health Organization (WHO) classification (17), and found 958 dengue sequences from Djibouti, Israel, Pakistan, Saudi Arabia, Somalia, and the United Arab Emirates. The sequence length parameter was set to 9,000 base pairs (bp) to take only complete and near-complete genomes between 10,000 and 11,000 (bp) to ensure analytical consistency and resolution across comparative analyses. Then, we selected sequences from 2013 onward, reflecting a critical period in dengue virus evolution following the emergence of the DENV-5 lineage first reported in 2013 (4). Sequences containing ambiguous nucleotides or large unresolved gaps were corrected or removed entirely. In rare cases where a key sequence could not be fully resolved but was critical for regional representation or lineage continuity, we substituted closely matched, high-quality sequences. For example, the Indian isolate (PV476196.1) was substituted with (OR259174.1), a cleaner representative from the same outbreak and genomic cluster. We also included the main 4 types of reference sequences (RefSeq) in the NCBI virus database, leading to 30 sequences to use. In addition, sequences from Cuba were included due to their role in the emergence of new dengue lineages, along with strains from India, China, Thailand, and Mali based on their geographic proximity and relevance to regional transmission dynamics; these additions help fill gaps caused by limited sequence availability across parts of the EMRO. Finally, a total of 47 sequences were used for further analysis. All sequences analyzed in this study were retrieved from publicly available records in the NCBI Virus database available at the time of data collection. The years reported throughout the manuscript (including 2025) refer to the collection year or database annotation associated with each sequence record. The final dataset used in this study is a retrospective comparative genomic study of the publicly available EMRO and non-EMRO strains used to illustrate the regional cluster findings.

8.2 Software and programs

Our results were produced using a range of software tools. MAFFT v7 was first used for sequence alignment to perform a general clean-up and organize the sequences before phylogenetic analysis, followed by re-alignment in MEGA11 using the MUSCLE algorithm with default parameters. We applied a gap open penalty of −400.00 and a gap extension of 0.00, with UPGMA used for clustering in both iterations; this approach ensured speed and accuracy for multiple sequence alignment across viral datasets (25, 39). All phylogenetic analyses were performed in MEGA11, where the nucleotide tree was constructed using the Tamura-Nei model under the Neighbor-Joining method with 1000 bootstrap replicates; the model incorporated a gamma distribution (G = 5.0) to account for evolutionary rate variation among sites with pairwise deletion used to handle gaps (40, 41). Codon positions 1st, 2nd, 3rd, and noncoding regions were all included in the tree-building process. The dataset was trimmed for amino acid-based analyses to include only the envelope (E) protein, removing all non-relevant regions. The remaining sequences were codon-aligned using MUSCLE in MEGA11 under default protein alignment settings (gap open = −2.90; hydrophobicity multiplier = 1.20). The resulting amino acid phylogenetic tree was built using the Neighbor-Joining method with the p-distance model to ensure direct and straightforward evolutionary distance measurements that calculate the proportion of nucleotide differences between sequences (21), also using gamma distribution (G = 5.0) with gaps treated by pairwise deletion, and bootstrap analysis repeated 1000 times; trees were rooted using the outgroup method to clarify directional divergence across strains (42). Lastly, we generated a molecular clock tree using the RelTime method to evaluate divergence timing and possible temporal clustering. We chose the China 2025 sequence as the outgroup based on its clear phylogenetic distance from all other strains. Sampling years were extracted from sequence names and assigned as tip dates, allowing MEGA to infer divergence times concerning known collection points and enabling a temporal view of the virus evolution in the dataset (43, 44).

8.3 Protein Level Analysis

The same dataset used for the nucleotide phylogenetic tree was translated into amino acid sequences using MEGA11; however, we added a newly reported DENV-3 Cuba strain, suspected to represent a distinct lineage, to evaluate whether it showed any relation to sequences in our dataset. After translation, amino acid substitutions were manually extracted from the alignments and documented. Check Table 1 for a summary of the sequences used and Supplementary Table 1 for each substitution point that was labeled with its position and categorized by severity. The finalized FASTA file of the translated protein sequences was then used to generate a Principal Component Analysis (PCA) plot and a mutation severity heatmap, both created using Python pipelines. The custom Python workflow was used only for downstream visualization and comparative feature extraction after sequence preprocessing and alignment. Specifically, translated E protein FASTA sequences were converted into numerical feature vectors using a k-mer frequency–based encoding approach, which represents each sequence as a fixed-length numerical vector based on amino acid subsequence frequencies (45). Subsequently for the heatmap, the classification was based on physicochemical differences between amino acids, including polarity, charge, and side-chain properties, which are widely used indicators of the potential structural and functional impact of protein substitutions in computational protein analysis.