LAUSR

Abstract DNA methylation is an elementary epigenetic process. The N6-methyladenine is related to a large kind of biological processes i.e., transcription, DNA replication, and repair. In genome, the N6-methyladenine (6 mA) site distribution is non-random; therefore, precise discrimination of 6 mA is necessary to understand its biological functions. Through biochemical experiments, the N6-methyladenine produced a positive outcome, still, these wet lab processes are very time consuming and high pricy. In view of this, it is of high priority to introduce a powerful, accurate, and fast computational model to identify N6-methyladenine sites. In this connection, we propose an intelligent computational model called iDNA6mA (5-step rule) using deep learning approach to identify N6-methyladenine sites from DNA sequences in the rice genome. Existing methods used handcrafted features to identify N6-methyladenine sites; however, the proposed computational model automatically extracts the key features from DNA input sequences via the proposed convolution neural network (CNN) model. The intelligent computational model iDNA6mA (5-step rule) obtained 86.64% of accuracy, 86.70% of sensitivity, 86.59% of specificity, 0.732 of MCC, and 0.931 of auROC. The results demonstrate that the proposed intelligent computational model achieved better performance in terms of all evaluation parameters than existing techniques. It is observed that iDNA6mA (5-step rule) model will become a useful tool in the fields of computational biology, bioinformatics, and for the academic research on N6-methyladenine sites prediction. A user-friendly webserver has been established and freely accessible at https://home.jbnu.ac.kr/NSCL/iDNA6mA.htm .