Genomics analysis of local Dengue virus to predict emergence of its novel serotype in Pakistan

Background: Dengue, which infects a huge population and causes mortality in the South Asian region, continues to be a major problem to combat. Dengue fever persists despite preventative measures being implemented. To safeguard highly populated areas, it is critical to stay one step ahead of the virus by forecasting evolutionary changes that may be responsible for its pathogenicity and severity. Methods: In this research, a genomic analysis of the local isolates of Dengue has been done to calculate the new emerging strain in the environment. For this study, a classification model has been designed to classify dengue serotypes by using a feed-forward neural network. The whole genome sequences of Pakistani isolates of dengue are retrieved from different databases. The unique characters have been extracted and converted into integers for ease of use in the deep learning model and to facilitate the process. The data set is divided into two sets: testing and training set which the model uses to train itself and then test itself.Results: The model has shown flawless performance with 100% accuracy. Different tests were conducted to assess the efficiency and performance of the model. Additionally, the model has exhibited competence in predicting novel serotypes based on pairwise sequence similarity.Conclusion: All the tests show that the multi-class classification model is accurate and correctly predicted and classified unknown serotypes.