Related Experiment Video
Updated: Aug 13, 2025
07:15
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
6.9K
CNN6mA: Interpretable neural network model based on position-specific CNN and cross-interactive network for 6mA site
Sho Tsukiyama1, Md Mehedi Hasan2, Hiroyuki Kurata1
1Department of Bioscience and Bioinformatics, Kyushu Institute of Technology, 680-4 Kawazu, Iizuka, Fukuoka 820-8502, Japan.
Computational and Structural Biotechnology Journal
|January 20, 2023
Summary
N6-methyladenine (6mA) is crucial for DNA processes and cancer. A new CNN-based predictor, CNN6mA, accurately identifies 6mA sites, overcoming limitations of traditional methods.
Area of Science:
- Epigenetics and Molecular Biology
- Bioinformatics and Computational Biology
- Genomics and DNA Methylation
Background:
- N6-methyladenine (6mA) is a vital epigenetic modification involved in DNA replication, repair, transcription, and disease, including cancer.
- Current methods for detecting 6mA sites genome-wide are often time-consuming and labor-intensive, necessitating more efficient approaches.
Purpose of the Study:
- To develop a novel computational tool for accurate and efficient prediction of N6-methyladenine (6mA) sites.
- To introduce innovative deep learning architectures for enhanced epigenetic analysis.
Main Methods:
- Development of CNN6mA, a Convolutional Neural Network (CNN)-based predictor for 6mA sites.
- Implementation of a position-specific 1-D convolutional layer to capture sequence-specific features at different levels.
- Design of a cross-interactive network to analyze nucleotide pattern relationships within DNA sequences.
Main Results:
- CNN6mA demonstrated superior performance compared to existing state-of-the-art models across multiple species.
- The predictor generated a contribution score vector, providing interpretable insights into the prediction mechanism.
- The tool offers a more efficient alternative to experimental methods for 6mA site detection.
Conclusions:
- CNN6mA provides an accurate and efficient method for predicting N6-methyladenine sites, advancing epigenetic research.
- The novel deep learning architectures contribute to the development of advanced bioinformatics tools for genomic analysis.
- The open accessibility of CNN6mA's source code and web application facilitates its adoption in the scientific community.
Keywords:
6mA, N6-methyladenineAUCs, Area under the curvesBERT, Bidirectional Encoder Representations from TransformersCNNCNN, Convolutional neural networkDNA modificationDeep learningInterpretable predictionLSTM, Long short-term memoryMCC, Matthews correlation coefficientMachine learningN6-methyladenineRF, Random forestSMRT, Single-molecule real-timeSN, SensitivitySP, SpecificityUMAP, Uniform manifold approximation and projectiont-SNE, t-distributed stochastic neighbor embedding
