Post-transcriptional modification such as N6-methyladenosine (m6A) has a crucial role in the stability and regulation of gene expression. Therefore, the identification of m6A is highly required for understanding the functional mechanisms of biological processes. Several machine learning techniques based on handy craft feature extraction methods have been proposed to facilitate the laborious work. However, due to the inefficient feature extraction, these techniques increase the computational complexity and thereby affect the identification accuracy of m6A.This paper proposes a fast and reliable predictive model for the identification of m6A sites. The proposed model is based on the convolutional neural network (CNN) which extracts the most significant features from the RNA sequences encoded by concatenating one-hot and nucleotide chemical properties. The proposed model is applied and tested on multiple species benchmark datasets and evaluated against the state-of-art predictive models. The results indicate that the proposed model achieves high accuracy of 93.6 %, 93.8 %, 85.0 % and 92.5 % on the datasets of Homo sapiens (H.sapien), Mus musculus (M.musculs), Saccharomyces cerevisiae (S.cerevisiae), and Arabidopsis thaliana (A.thaliana), respectively.The proposed model could be used to facilitate the researcher’s community in m6A identification. In addition, an easy to use web server is made available at https://home.jbnu.ac.kr/NSCL/pm6acnn.htm.

Piwi-interacting RNAs (piRNAs) are a distinct sub-class of small non-coding RNAs that are mainly responsible for germline stem cell maintenance, gene stability, and maintaining genome integrity by repression of transposable elements. piRNAs are also expressed aberrantly and associated with various kinds of cancers. To identify piRNAs and their role in guiding target mRNA deadenylation, the currently available computational methods require urgent improvements in performance. To facilitate this, we propose a robust predictor based on a lightweight and simplified deep learning architecture using a convolutional neural network (CNN) to extract significant features from raw RNA sequences without the need for more customized features. The proposed model’s performance is comprehensively evaluated using $k$ -fold cross-validation on a benchmark dataset. The proposed model significantly outperforms existing computational methods in the prediction of piRNAs and their role in target mRNA deadenylation. In addition, a user-friendly and publicly-accessible web server is available at http://nsclbio.jbnu.ac.kr/tools/2S-piRCNN/ .

Promoters are important regulatory elements in the genome that control gene expression, and their abnormalities have been linked to various diseases. With the development of high-throughput sequencing techniques, there is a need for computational methods to identify promoters in large amounts of DNA sequence data. However, many existing methods rely on a single feature representation approach, which can potentially lead to information loss. In order to address this issue. We proposed a computational model iPro-TCN that combines Temporal Convolutional Network (TCN) with a word2vec feature representation. This model includes a new feature descriptor called K-mer word vector and has been shown to have high sensitivity and accuracy in distinguishing promoters, including strong and weak promoters. The iPro-TCN model was evaluated on a benchmark dataset and was able to achieve good performance in both promoter identification and promoter strength prediction. Specifically, the model had an accuracy of 91.86% in the first layer for promoter identification, and an accuracy of 84.63% in the second layer for promoter strength prediction. These results suggest that the iPro-TCN model is a strong performer in predicting promoter sequences and promoter strength in DNA. These results are far higher than the existing best predictor, which indicates that our model showcases a better performance compared with the existing approaches.

DNA is subject to epigenetic modification by the molecule N4-methylcytosine (4mC). N4-methylcytosine plays a crucial role in DNA repair and replication, protects host DNA from degradation, and regulates DNA expression. However, though current experimental techniques can identify 4mC sites, such techniques are expensive and laborious. Therefore, computational tools that can predict 4mC sites would be very useful for understanding the biological mechanism of this vital type of DNA modification. Conventional machine-learning-based methods rely on hand-crafted features, but the new method saves time and computational cost by making use of learned features instead. In this study, we propose i4mC-Deep, an intelligent predictor based on a convolutional neural network (CNN) that predicts 4mC modification sites in DNA samples. The CNN is capable of automatically extracting important features from input samples during training. Nucleotide chemical properties and nucleotide density, which together represent a DNA sequence, act as CNN input data. The outcome of the proposed method outperforms several state-of-the-art predictors. When i4mC-Deep was used to analyze G. subterruneus DNA, the accuracy of the results was improved by 3.9% and MCC increased by 10.5% compared to a conventional predictor.

The 5-Methyluridine (m5U), predominantly present in RNA and especially enriched in transfer RNA (tRNA), significantly enhances translational accuracy and protein synthesis by ensuring precise genetic information decoding and optimal tRNA functionality within cellular mechanisms. The identification of m5U modification sites is crucial, as this modification has gained significant attention in diseases such as breast cancer, stress response, and viral infections, offering insights into its molecular mechanisms and regulatory functions in disease contexts. Nevertheless, due to the arduous nature, intricate procedures, reliance on sophisticated and expensive instrumentation, and the need for specialized expertise, conventional biochemical approaches for identifying m5U modification sites result in substantial resource expenditures and notable temporal investments. Consequently, the pressing need for a precise and efficient computational method highlights the urgency for alternative approaches in identifying m5U modification sites. In this study, we introduce a novel computational approach called “Deep-m5U,” which combines the strengths of Convolutional Neural Networks (CNNs) and tetranucleotide composition to accurately identify methyluridine modification sites and improve overall performance. The developed Deep-m5U method leverages CNNs to accurately detect protein-coding regions aand capture relevant motifs, while incorporating tetra-nucleotide composition to capture global compositional characteristics, resulting in a more robust model that significantly enhances performance. We evaluated the Deep-m5U model on two publicly available benchmark datasets: the full transcript and mature mRNA datasets. Our results showcase superior performance, achieving accuracies of 91.26% and 95.63% respectively, surpassing the current cutting-edge methods. Moreover, the open-source code for Deep-m5U is freely accessible at: https://github.com/waleed551/Deep-m5U .