methimpute

methimpute implements HMM-based imputation to infer methylation status and methylation levels for individual cytosines from whole-genome bisulfite sequencing (WGBS) data, enabling reconstruction of complete, cytosine-resolution plant methylomes from sparse coverage.

Key Features:

• Hidden Markov Model (HMM) imputation: Uses an HMM-based algorithm to infer methylation states across genomic cytosines.
• Cytosine-resolution inference: Infers both methylation status and quantitative methylation level for each cytosine irrespective of initial coverage.
• Low-coverage reconstruction: Can generate complete methylomes from low-coverage WGBS data, reported down to ~6X sequencing depth.
• Reduced depth requirement: Enables methylome reconstruction at substantially lower depth compared to typical requirements around 60X.
• Validation on plant species: Algorithm performance has been evaluated on maize, rice, and Arabidopsis with high accuracy.
• Applicability across species: Demonstrated utility in plants and reported potential applicability to a broad range of other species.
• Works with WGBS base-resolution data: Operates on whole-genome bisulfite sequencing inputs to recover base-resolution methylation profiles.

Scientific Applications:

• Complete methylome reconstruction: Reconstruct genome-wide, cytosine-resolution methylomes from sparse WGBS datasets.
• Population-scale and large-genome studies: Facilitate comparative and population epigenomic analyses when deep sequencing is cost-prohibitive.
• Plant epigenomics: Support methylation analysis in plants, including maize, rice, and Arabidopsis.
• Cross-species methylation analysis: Enable exploration of DNA methylation patterns across diverse species where coverage is limited.

Methodology:

Imputation is performed using a Hidden Markov Model to infer per-cytosine methylation status and methylation level from WGBS data.