ParRADMeth

ParRADMeth identifies differentially methylated (DM) regions from large methylation count datasets using a parallelized beta-binomial regression to enable scalable, statistically robust analysis.

Key Features:

• Parallel Processing: Leverages parallel computing across multicore CPU clusters to reduce runtime for identifying DM regions.
• Beta-Binomial Regression: Employs beta-binomial regression to model overdispersed methylation count data.
• High Computational Efficiency: Demonstrated up to 189 times faster performance than sequential approaches on a cluster of 16 nodes, each equipped with two eight-core processors.
• Biological Accuracy: Builds upon the sequential RADMeth method and maintains the validated biological accuracy reported in experimental evaluations.

Scientific Applications:

• Large-scale epigenetic studies: Identification of DM regions in large methylation datasets for research into epigenetic mechanisms underlying disease processes, including cancer, neurological disorders, and other conditions influenced by DNA methylation patterns.

Methodology:

Parallelized beta-binomial regression applied to methylation count data across multiple cores in a CPU cluster.