iN6-methylat (5-step): identifying DNA N-6-methyladenine sites in rice genome using continuous bag of nucleobases via Chou's 5-step rule

摘要

DNA N-6-methyladenine is a non-canonical DNA modification that occurs in different eukaryotes at low levels and it has been identified as an extremely important function of life. Moreover, about 0.2% of adenines are marked by DNA N-6-methyladenine in the rice genome, higher than in most of the other species. Therefore, the identification of them has become a very important area of study, especially in biological research. Despite the few computational tools employed to address this problem, there still requires a lot of efforts to improve their performance results. In this study, we treat DNA sequences by the continuous bags of nucleobases, including sub-word information of its biological words, which then serve as features to be fed into a support vector machine algorithm to identify them. Our model which uses this hybrid approach could identify DNA N-6-methyladenine sites with achieved a jackknife test sensitivity of 86.48%, specificity of 89.09%, accuracy of 87.78%, and MCC of 0.756. Compared to the state-of-the-art predictor as well as the other methods, our proposed model is able to yield superior performance in all the metrics. Moreover, this study provides a basis for further research that can enrich a field of applying natural language-processing techniques in biological sequences.