Geographic Classification and Identification of SARS-CoV2 From Related Viral Sequences

Abstract

The COVID-19 pandemic has introduced to mild the risks of deadly epidemic-prone illnesses sweeping our globalized planet. The pandemic is still going strong, with additional viral variations popping up all the time. For the close to future, the international response will have to continue. The molecular tests for SARS-CoV-2 detection may lead to False-negative results due to their genetic similarity with other coronaviruses, as well as their ability to mutate and evolve. Furthermore, the clinical features caused by SARS-CoV-2 seem to be like the symptoms of other viral infections, making identification even harder. We constructed seven hidden Markov models for each coronavirus family (SARS-CoV2, HCoV-OC43, HCoV-229E, HCoV-NL63, HCoV-HKU1, MERS-CoV, and SARS-CoV), using their complete genome to accurate diagnose human infections. Besides, this study characterized and classified the SARS-CoV2 strains according to their different geographical regions. We built six SARS-CoV2 classifiers for each world's continent (Africa, Asia, Europe, North America, South America, and Australia). The dataset used was retrieved from the NCBI virus database. The classification accuracy of these models achieves 100% in differentiating any virus model among others in the Coronavirus family. However, the accuracy of the continent models showed a variable range of accuracies, sensitivity, and specificity due to heterogeneous evolutional paths among strains from 27 countries. South America model was the highest accurate model compared to the other geographical models. This finding has vital implications for the management of COVID-19 and the improvement of vaccines.