SOMNiBUS

SOMNiBUS analyzes count-based methylation data on predefined genomic regions to identify differentially methylated regions associated with phenotypes or traits.

Key Features:

• Smooth covariate effects: The model allows effects of multiple covariates on methylation levels to vary smoothly along genomic regions.
• Sequencing error handling: The method explicitly allows for experimental errors in bisulfite sequencing data.
• Cell type mixture adjustment: The method can adjust for variability in cell type mixture as a confounding factor.
• Targeted sequencing compatibility: The package is designed for data from targeted sequencing approaches with predefined genomic regions.

Scientific Applications:

• Differential methylation analysis: Identification of differentially methylated regions associated with phenotypes or traits in count-based methylation data.
• Disease-associated methylation studies: Detection of disease-associated changes in DNA methylation, as demonstrated in rheumatoid arthritis patients and controls.

Methodology:

SOMNiBUS implements a regression-based hierarchical model with a specialized EM algorithm for inference about smooth covariate effects.