DamMet: ancient methylome mapping accounting for errors, true variants, and post-mortem DNA damage

Area of Science:

• Ancient DNA research
• Genomics
• Bioinformatics

Background:

• Ancient DNA (aDNA) research increasingly utilizes genome-wide methylation patterns.
• Current methods for inferring DNA methylation from aDNA are limited by confounding factors like sequencing errors and true variants.
• Variability in methylation measurements across samples hinders direct comparisons.

Purpose of the Study:

• To introduce DamMet, an open-source software for estimating DNA methylation levels from aDNA.
• To develop a novel statistical model that accounts for post-mortem DNA damage, sequencing errors, and varying deamination rates.
• To provide a reliable and cross-sample comparable method for analyzing ancient DNA methylation.

Main Methods:

• Developed DamMet, a C++ software utilizing a maximum likelihood approach.
• Created a novel statistical model for dinucleotide post-mortem DNA damage.
• Extended the gargammel ancient DNA simulator to include methylation-dependent decay features for validation.

Main Results:

• DamMet provides accurate maximum likelihood estimates of regional CpG methylation levels in aDNA.
• The software accounts for sequencing errors, genotypes, and differential cytosine deamination rates.
• DamMet's inferred methylation levels correlate with existing tools (epiPALEOMIX) and are comparable to fresh tissue bisulfite sequencing data.

Conclusions:

• DamMet offers genuine, cross-sample comparable estimates of local DNA methylation in ancient genomes.
• The open-source software is available with documentation for broader accessibility in ancient genomics research.