Related Experiment Video
Updated: Aug 26, 2025
13:47
Enhanced Reduced Representation Bisulfite Sequencing for Assessment of DNA Methylation at Base Pair Resolution
Published on: February 24, 2015
25.7K
MSNet-4mC: learning effective multi-scale representations for identifying DNA N4-methylcytosine sites
Chunting Liu1,2, Jiangning Song3,4, Hiroyuki Ogata2
1Department of Intelligence Science and Technology, Graduate School of Informatics, Kyoto University, Kyoto, Kyoto 606-8501, Japan.
Bioinformatics (Oxford, England)
|October 7, 2022
Summary
We developed MSNet-4mC, a computational method to identify N4-methylcytosine (4mC) sites in DNA. This approach significantly improves prediction accuracy over existing methods.
Area of Science:
- Genomics
- Epigenetics
- Bioinformatics
Background:
- N4-methylcytosine (4mC) is a critical epigenetic modification regulating biological processes.
- Experimental detection of 4mC sites is laborious and time-consuming.
- Computational methods offer an efficient alternative for identifying 4mC sites.
Purpose of the Study:
- To develop an effective computational method for identifying 4mC sites.
- To improve the predictive capability by exploiting complex DNA sequence interactions.
Main Methods:
- Proposed MSNet-4mC, a lightweight neural network utilizing convolutional operations.
- Employed multi-scale receptive fields to capture short and long-range DNA sequence relationships.
- Applied class weights in cross-entropy loss to address data imbalance across species.
Main Results:
- MSNet-4mC demonstrated significant performance improvements in identifying 4mC sites.
- The method outperformed existing state-of-the-art computational approaches.
- Benchmarking experiments confirmed the efficacy of the proposed network architecture.
Conclusions:
- MSNet-4mC provides an accurate and efficient computational tool for 4mC site identification.
- The method effectively leverages sequence information for improved epigenetic analysis.
- The developed tool is freely available for research use.
