Water cycle tunicate swarm algorithm based deep residual network for virus detection with gene expression data

ABSTRACT

The virus is pervasive in the environment. The risk of pathogens relies on the fact that several viruses can quickly mutate with diagnostic measures. The measures of gene expression data have proved to be a beneficial application for medical treatment and for examining basic biology. This study uses gene expression data to design a deep architecture driven by optimisation for viral detection. The whole procedure taken into account in the created model includes feature selection, data transformation, and virus detection. The initial step is to gather gene expression data. Yeo-Jhonson transformation is used to change the structure and format of raw data during data transformation. After the data has been transformed, mutual information is used to choose the features. Then, a deep residual network (DRN) is constructed using the Water Cycle Tunicate Swarm Algorithm (WCTSA), which is used to detect viruses as the last step. Combining the Water Cycle Algorithm (WCA) and the Tunicate Swarm Algorithm (TSA) led to the development of the WCTSA. With the greatest accuracy of 91.4%, sensitivity of 92.6%, and specificity of 90.7%, the proposed approach offered improved efficiency.

KEYWORDS:

• Virus detection
• gene expression data
• deep residual network
• Yeo-Johnson transformation
• mutual information

Disclosure statement

No potential conflict of interest was reported by the authors.