Systematic Literature Review: Virus Prediction Based on DNA Sequences using Machine Learning and Deep Learning method

Abstract

Many methods used to do viruses prediction both with Machine Learning and Deep Learning algorithms. By using these methods, DNA sequence data can be categorized more efficient and easily. Furthermore, the evaluation result from it can be reviewed for future research. In general, systematic review can be used to analyze trends and identify several things such as the methods, topics related, dataset used and also provide answer regarding virus prediction based on its DNA sequences. This research chooses and analyze 31 papers from related studies on this field between 2015–2022. From selected papers it found that in general there are 5 main scopes and processes to predict viruses which are Feature Extraction, Distance Counting, Clustering, Classification, and Evaluation Metrics. The dataset also indicates that total 79% of selected paper were using the same datasets from NCBI official database. Furthermore, hybrid model of K-means offered high evaluation metrics result and be used for future research. Deep Learning approach which are Convolutional Neural Network and its hybrid model such as CNN-Bi-LSTM can also be used because of its high accuracy and performance quality which exceed 90% to classify viruses on various studies and almost 100% at its peak.