Utilizing chaos game representation for enhanced classification of SARS-CoV-2 variants with stacked sparse autoencoders.

Since the beginning of the COVID-19 pandemic, the World Health Organization (WHO) has been tracking SARS-CoV-2 mutations. The SARS-CoV-2 consistently mutated throughout the pandemic, which resulted in many variants. A variant is a viral genome containing one or more genetic code mutations. Deep learning techniques have been successfully used in many viral classification problems associated with viral infection diagnosis, metagenomics, phylogenetics, and analysis. This work proposed an effective viral genome classifier for SARS-CoV-2 variants using the deep neural network based on the stacked sparse autoencoder (SSAE). Aiming to achieve the best performance of the model, we explored the utilization of image representations of the complete genome sequences as the SSAE input. The dataset based on Chaos Game Representation (CGR) images was generated and applied to the experiments of classification of SARS-CoV-2 variants of concern (VOC). The SSAE technique provided great performance results, achieving classification accuracy of 99.9% for the validation set and 99.8% for the test set. Finally, the results indicated the relevance of using this deep learning technique in genome classification problems. [ABSTRACT FROM AUTHOR]