HOME

HOME identifies differentially methylated regions (DMRs) from whole-genome bisulfite sequencing data using histogram-of-methylation features and Support Vector Machine (SVM) classification to detect methylation differences at single-base resolution.

Key Features:

• Histogram Of MEthylation features: Computes histogram-based features that capture methylation level distributions distinguishing DMRs from non-DMRs.
• Support Vector Machine classification: Uses an SVM classifier to separate DMRs and non-DMRs based on extracted features.
• Dataset-independent feature generation: Generates features intended to be transferable across organisms, enabling training on one methylome and application to another.
• Whole-genome bisulfite sequencing (WGBS) single-base resolution: Operates on single-base methylation measurements derived from WGBS data.
• Non-CG context analysis: Detects DMRs in non-CG methylation contexts in addition to CpG contexts.
• Time series analysis: Supports analysis of methylation dynamics across time-series datasets.
• Improved accuracy and boundary identification: Reduces false positives and refines DMR boundaries compared to alternative approaches.
• Biological relevance linkage: Identifies DMRs that associate with genes, biological processes, and regulatory events.

Scientific Applications:

• DMR detection: Identification of differentially methylated regions from WGBS data at single-base resolution.
• Methylation dynamics studies: Analysis of temporal methylation changes using time-series data.
• Non-CG methylation analysis: Examination of methylation patterns in non-CG contexts across genomes.
• Cross-species methylome analysis: Application of trained classifiers across different organisms to detect conserved or species-specific DMRs.
• Integration with gene regulation studies: Linking identified DMRs to genes, regulatory events, and biological processes.

Methodology:

HOME computes histogram-of-methylation features from whole-genome bisulfite sequencing single-base methylation data and classifies regions using a Support Vector Machine trained on dataset-independent features.