RoAM: computational reconstruction of ancient methylomes and identification of differentially methylated regions

Abstract

Identifying evolutionary changes in DNA methylation bears a huge potential for unraveling adaptations that have occurred in modern humans. Over the past decade, computational methods to reconstruct DNA methylation patterns from ancient DNA sequences have been developed, allowing for the exploration of DNA methylation changes during the past hundreds of thousands of years of human evolution. Here, we introduce a new version of RoAM (Reconstruction of Ancient Methylation), a flexible tool that allows for the reconstruction of ancient methylomes, as well as the identification of differentially methylated regions between ancient populations. RoAM incorporates a series of filtering and quality control steps, resulting in highly reliable DNA methylation maps that exhibit similar characteristics to modern maps. To showcase RoAM’s capabilities, we used it to compare ancient methylation patterns between pre- and post-Neolithic revolution samples from the Balkans. Differentially methylated regions separating these populations are shown to be associated with genes related to regulation of sugar metabolism. Notably, we provide evidence for overexpression of the gene PTPRN2 in post-Neolithic revolution samples. PTPRN2 is a key regulator of insulin secretion, and our finding is compatible with hypoinsulinism in pre-Neolithic revolution hunter-gatherers. Additionally, we observe methylation changes in the genes EIF2AK4 and SLC2A5, which provide further evidence to metabolic adaptations to a changing diet during the Neolithic transition. RoAM offers powerful algorithms that position it as a key asset for researchers seeking to identify evolutionary regulatory changes through the lens of paleoepigenetics.