HPG-DHunter

HPG-DHunter detects differentially methylated regions (DMRs) in DNA methylation data by applying Discrete Wavelet Transform (DWT) signal analysis to enable rapid, genome-scale identification of methylation differences.

Key Features:

• Wavelet-based DMR detection: Uses Discrete Wavelet Transform (DWT) to compare methylation signals at multiple resolution levels for automatic identification of DMRs.
• Multi-scale visualization: Produces visual representations of DMRs across scales from local regions to genome-wide contexts.
• GPU acceleration: Leverages NVIDIA GPUs and the CUDA programming model to accelerate computation on large methylation datasets.
• Integration with HPG-MSuite: Integrates into HPG-MSuite workflows for DNA alignment, methylation analysis, and generation of genome-wide methylation maps.
• Scalability and performance: Validated to process large datasets (benchmark: twelve human chromosomes across six samples, aggregated ~108 GB) completing DMR detection in approximately 3.5 hours with concordance comparable to other tools.
• Dataset compatibility: Demonstrates similar concordance when analyzing both real and synthetic methylation datasets.

Scientific Applications:

• Epigenetics research: Identification of DMRs associated with diseases, developmental processes, or environmental influences.
• Genome-wide methylation mapping: Generation and analysis of methylation maps across multiple human chromosomes.
• Method benchmarking and validation: Comparative evaluation of DMR detection methods using real and synthetic datasets.

Methodology:

Transforms methylation signals using Discrete Wavelet Transform (DWT) to compare signal resolutions for DMR detection and employs NVIDIA GPU acceleration via the CUDA programming model.