MLML2R

MLML2R estimates maximum likelihood proportions of 5-methylcytosine (5-mC) and 5-hydroxymethylcytosine (5-hmC) at single-nucleotide resolution by integrating data from bisulfite sequencing (BS), oxidative bisulfite sequencing (oxBS), and Tet-Assisted bisulfite conversion (TAB).

Key Features:

• Maximum Likelihood Estimation: Provides maximum likelihood estimates (MLE) of 5-mC and 5-hmC proportions at single-nucleotide resolution.
• Multi-method Integration: Integrates measurements from bisulfite sequencing (BS), oxidative bisulfite sequencing (oxBS), and Tet-Assisted bisulfite conversion (TAB) to resolve methylation and hydroxymethylation.
• Analytical two-method solutions: Derives exact constrained MLEs for any two-method combination (e.g., BS+oxBS, BS+TAB) using the pool-adjacent-violators algorithm (PAVA).
• Three-method solutions: Implements an iterative estimator based on Qu et al. (2013) and a non-iterative Lagrange multiplier approximation for BS+oxBS+TAB, with the Lagrange approach reducing computational time.
• Input data flexibility: Accepts preprocessed intensities from Infinium Methylation arrays and counts from Next Generation Sequencing technologies.

Scientific Applications:

• Gene regulation studies: Quantification of DNA methylation and hydroxymethylation to investigate regulatory mechanisms.
• Cancer biology: Analysis of methylation and hydroxymethylation alterations in cancer research.
• Developmental biology: Characterization of epigenetic dynamics during developmental processes.

Methodology:

Computational methods include maximum likelihood estimation using the pool-adjacent-violators algorithm (PAVA) for two-method combinations, an iterative estimator based on Qu et al. (2013) and a non-iterative Lagrange multiplier approximation for three-method combinations, with input as Infinium array intensities or NGS counts.