REMP

REMP predicts locus-specific DNA methylation levels of repetitive elements using a random forest algorithm to extend Illumina Infinium methylation array data for high-resolution genome-wide and single-base epigenetic profiling.

Key Features:

• Random forest modelling: Employs a random forest algorithm to capture non-linear relationships and predict methylation status across repetitive element loci.
• Integration with Infinium arrays: Uses Illumina Infinium methylation array profiling data to infer locus-specific methylation of repetitive elements despite limited CpG coverage on arrays.
• High-resolution outputs: Produces predictions at genome-wide and single-base resolution for repetitive elements.
• Feature integration: Leverages surrounding genetic and epigenetic information as predictors for locus-specific methylation.
• Validation: Predictive performance has been validated using alternative sequencing and microarray datasets.
• Large-cohort application: Extends locus-specific repetitive element methylation information to large datasets such as The Cancer Genome Atlas (TCGA).

Scientific Applications:

• Epigenome-wide association studies (EWAS): Enables inclusion of repetitive element methylation profiles in EWAS to increase resolution and specificity.
• Differentially methylated region (DMR) analysis: Facilitates identification of DMRs within repetitive elements for studies of epigenetic regulation.
• Cancer research: Supports analysis of repetitive element methylation changes observed in tumors and their role in cancer development and progression.

Methodology:

A random forest model is trained on Illumina Infinium methylation array profiling data, leveraging surrounding genetic and epigenetic features to predict methylation levels at specific loci within repetitive elements; predictive performance was validated using alternative sequencing and microarray data.