Related Experiment Video
Updated: Mar 14, 2026
08:12
A Web-Based Workflow for Selecting Gene- and Tissue-Specific Enhancers
Published on: July 18, 2025
775
LMethyR-SVM: Predict Human Enhancers Using Low Methylated Regions based on Weighted Support Vector Machines
Jingting Xu1, Hong Hu1, Yang Dai1
1Department of Bioengineering, University of Illinois at Chicago, Chicago, Illinois, United States of America.
Plos One
|September 24, 2016
Summary
This study introduces LMethyR-SVM, a novel framework using DNA methylation profiles to predict cell-type-specific enhancers. The method effectively identifies enhancers, complementing existing epigenetic data for improved regulatory region discovery.
Area of Science:
- Genomics and Epigenetics
- Computational Biology
- Regulatory Element Identification
Background:
- Enhancer identification is challenging, with current models relying on histone modifications.
- DNA methylation, particularly low methylated regions (LMRs), has been underexplored for enhancer prediction.
- LMRs are implicated as distal active regulatory regions.
Purpose of the Study:
- To develop a novel prediction framework, LMethyR-SVM, for cell-type-specific enhancers.
- To leverage whole genome bisulfite sequencing (WGBS) DNA methylation profiles for enhancer prediction.
- To explore the utility of LMRs in enhancer identification.
Main Methods:
- Utilized cell-type-specific WGBS DNA methylation profiles to identify LMRs.
- Developed a weighted support vector machine (SVM) learning framework (LMethyR-SVM).
- Classified LMRs into reliable positive, like positive, and likely negative sets based on validated enhancers from the VISTA database.
Main Results:
- Demonstrated LMethyR-SVM performance using WGBS data from human embryonic stem cells (H1) and fetal lung fibroblasts (IMR90).
- Predicted enhancers showed high conservation and reasonable validation rates with known markers (transcription factors, p300 binding, DNase-I sites).
- A significant fraction of LMethyR-SVM predicted enhancers were not identified by ChromHMM and were enriched for FANTOM5 enhancers.
Conclusions:
- Low methylated regions from WGBS data are valuable complementary resources for enhancer prediction.
- LMethyR-SVM enhances the prediction of cell-type-specific enhancers.
- This approach improves upon models relying solely on histone modification marks.
