MLML2R: an R package for maximum likelihood estimation of DNA methylation and hydroxymethylation proportions

Area of Science:

• Epigenetics
• Bioinformatics
• Computational Biology

Background:

• Accurate measurement of 5-methylcytosine (5-mC) and 5-hydroxymethylcytosine (5-hmC) at single-nucleotide resolution is crucial for understanding epigenetic regulation.
• Existing computational tools often have limitations in handling data from various DNA processing methods like traditional (BS), oxidative (oxBS), and Tet-Assisted (TAB) bisulfite conversion.

Purpose of the Study:

• To introduce MLML2R, an R package for calculating maximum likelihood estimates (MLE) of 5-mC and 5-hmC proportions.
• To provide a flexible and computationally efficient tool for analyzing epigenetic data from diverse bisulfite conversion strategies and platforms.

Main Methods:

• Development of the MLML2R R package implementing maximum likelihood estimation (MLE).
• Derivation of pool-adjacent-violators algorithm (PAVA) exact constrained MLE for two-method combinations.
• Implementation of iterative and a novel non-iterative approximation (Lagrange multipliers) for three-method combinations.
• Compatibility with Infinium Methylation arrays and Next Generation Sequencing (NGS) count data.

Main Results:

• MLML2R provides MLE for 5-mC and 5-hmC, supporting combinations beyond the oxBS+BS method.
• The novel non-iterative approach significantly reduces computational time for analyzing high-throughput epigenetic data.
• The package accommodates various DNA processing methods and data types, enhancing flexibility.

Conclusions:

• MLML2R offers a robust and efficient solution for precise quantification of DNA methylation and hydroxymethylation.
• The package facilitates advanced epigenetic analyses by integrating data from multiple bisulfite conversion techniques.
• MLML2R is a valuable resource for researchers working with large-scale epigenetic datasets, available on CRAN.