PANDA: Phased Analysis of DNA Amplicons for Read-Level Methylation Profiling and Heterogeneity Assessment

Abstract

Targeted bisulfite amplicon sequencing is widely used to examine methylation at defined genomic loci, but site-wise methylation estimates can obscure phased patterns among individual reads. We developed PANDA (Phased Analysis for DNA Amplicons), an open-source platform for read-level analysis of targeted bisulfite amplicons generated by Sanger sequencing or amplicon NGS, with graphical and command-line interfaces built on a shared computational core (https://github.com/kubo-azu/PANDA). PANDA integrates reference-coordinate methylation calling and visualization, motif-based haplotype filtering, coordinate linkage of unmerged paired reads, and group comparison. It also implements three complementary heterogeneity metrics: Amplicon PDR (proportion of discordant reads), Window Epipolymorphism, and Amplicon qFDRP (quantitative fraction of discordant read pairs). These metrics use retained read-count weighting and explicit eligibility criteria. Synthetic demonstration datasets illustrated global methylation estimation, sequence-defined haplotype selection, and CpG-level group comparison. The heterogeneity metrics illustrated different aspects of the predefined methylation states. Re-analysis of public chimpanzee and rhesus macaque placental datasets was consistent with the previously reported species-associated methylation contrast and demonstrated coordinate-linked analysis of unmerged paired reads without imputing the unsequenced interval. PANDA provides a reproducible framework for examining phased methylation patterns and within-sample heterogeneity alongside site-wise methylation estimates in amplicon sequencing.