LuxHMM: DNA methylation analysis with genome segmentation via hidden Markov model

Area of Science:

• Genomics
• Epigenetics
• Bioinformatics

Background:

• DNA methylation is crucial for epigenetics and disease research.
• Analyzing differentially methylated regions (DMRs) is more informative than single CpG analysis due to correlated methylation patterns.

Purpose of the Study:

• To develop a novel probabilistic method and software (LuxHMM) for identifying DMRs.
• To enable robust analysis of differential methylation while accounting for experimental factors.

Main Methods:

• Utilizes a hidden Markov model (HMM) for genome segmentation.
• Employs a Bayesian regression model for differential methylation inference, accommodating multiple covariates.
• Incorporates experimental parameters specific to bisulfite sequencing.
• Inference performed using variational inference or Hamiltonian Monte Carlo (HMC).

Main Results:

• LuxHMM effectively segments the genome into regions.
• The Bayesian model accurately infers differential methylation in regions.
• The software handles multiple covariates and experimental parameters.

Conclusions:

• LuxHMM demonstrates competitive performance against existing methods.
• The software is suitable for analyzing real and simulated bisulfite sequencing data.
• Provides a powerful tool for epigenetic research and disease studies.