4mCPred-Caps: Identification of DNA N4-methylcytosine sites using CapsuleNet

N4-methy1cytosine (4mC) is amongst the most significant DNA modifications and is associated with the progression of gene expression, gene differentiation, and cell proliferation. It is essential to locate their alteration sites in the genome sequences in order to understand the biological roles of 4mC. Deep learning has recently gained popularity and is widely used for the identification of 4mC sites. In this study, a convolutional neural network (CNN) based model using CapsuleNet, 4mCPred-Caps, was created to categorize 4mC sites in three different species, Arabidopsis thaliana, Caenorhabditis elegans, and Drosophila melanogaster. Using the benchmark datasets of these species, we assessed the model performance and equate it with the previously available methodologies. We built our model using a single encoding technique called one-hot-encoding along with a special architecture called CapsuleNet. The proposed model acquired an accuracy of 84.8%, 90%, and 88.2% respectively on Arabidopsis thaliana, Caenorhabditis elegans, and Drosophila melanogaster respectively. The outcomes demonstrate that the proposed model surpassed the currently available tools in terms of all assessment parameters. The results collected indicate that the suggested approach can be very beneficial in the area of bioinformatics.