rDNAcaller: a fast and robust pipeline to call ribosomal DNA variants

Ribosomal DNA (rDNA) is arranged in tandem repeats in the human genome. Extensive genetic and copy number variation in rDNA has been reported both within and between individuals, contributing to phenotypic diversity. However, traditional variant callers, designed for diploid regions, have not been systematically benchmarked for analyzing rDNA. We developed a customized simulator that replicates real intra- and inter-individual rDNA variation to benchmark the performance of commonly used variant callers, including GATK, Mutect2, and Lofreq*. Additionally, we optimized the preprocessing and mapping steps to significantly improve accuracy. Based on these optimizations, we introduce rDNAcaller, a novel pipeline for accurate rDNA variant detection using short-read whole-genome sequencing that achieves 94% precision on experimental data. Applying our pipeline to data from the 1000 Genomes Project, we identify 5607 novel rDNA variant positions across human populations, with African individuals showing the highest number of variants. Overall, rDNAcaller is a robust and versatile tool for analyzing rDNA variation, addressing the limitations of existing methods in handling high ploidies. By enabling accurate detection of rDNA variants, it facilitates deeper exploration of rDNA's role in phenotypic diversity, supporting future genomic studies and broadening our understanding of rDNA biology in health and disease.