Assessing Phasing Accuracy for Inferring Haplotype Structure in Laboratory Marmosets

Adon C. Roy

Document Type

Article

Publication Date

2025

Keywords

JMG

Abstract

To understand the genetics behind aging and complex diseases like Alzheimer’s, we need accurate ways to model how DNA varies between individuals. While studying genetic changes at singular positions (SNPs) is common, looking at haplotypes gives a deeper understanding of how genetic traits are inherited and organized. The common marmoset (Callithrix jacchus) is a useful model for studying these diseases due to similarities with humans, but broader genomic architecture including haplotype structure hasn’t been previously studied for this non-human primate species. In this project, we tested two phasing tools, Beagle and WhatsHap, to see which one more accurately identifies haplotypes from short-read whole-genome sequencing data. We used a high-confidence “truth set” created with WhatsHap, which incorporates pedigree structure (sequenced trios) with a read-backed algorithm to resolve haplotypes. We then compared this to two test sets: one phased with Beagle, which uses population-based statistical modeling through Hidden Markov Models (HMMs), and the other with WhatsHap without including any pedigree information. without including any pedigree information. WhatsHap with trio input performed best across all accuracy measures, including metrics like switch error rate, switch/flip rate and Hamming distance. Visual analysis with IGV showed that pedigree-based phasing better matched regions of strong linkage disequilibrium. Finally, we found that marmosets show LD decay patterns similar to humans, further supporting their value as a model for studying genetically complex traits. These results highlight the importance of using family data when phasing genomes in non-human primates and provide a strong foundation for future genetic studies in marmosets.

Please contact the Joan Staats Library for information regarding this document.

COinS