DNA sequence clustering and ERSIT-GRU for repeat detection in COVID-19 prediction

Abstract

CoV research is ongoing to learn more about the virus's genesis and evolution, which has facilitated the quick prediction of its sickness. Atypical genetic sequences—which might be singular in general—are investigated in order to accomplish this. However, it is still difficult to investigate the different kinds of palindromes, mirror, and inverted repeats. Thus, a CoV variations prediction model is proposed in this article. Initially, the Sliding Window (SW) method is used once the Deoxyribo Nucleic Acid (DNA) sequences are gathered. Next, the NWJ-K-Means technique is used to identify and group the various sequences in the data. For each group, the sequence Association score is calculated. E-NJA is used to build the genomic tree based on this score. Next, feature extraction is done, and the FS approach is used to choose the best features. In the meantime, one-hot encoding is used to encode the initially aligned data. Lastly, the ERSIT-GRU is trained with the encoded vector and the best features in order to predict the CoV variations. During the experimental assessment, it was discovered that the suggested system was more effective than the current techniques.