Pairwise Paternity Assignment With Forward-Backward Simulations: Refining CERVUS Using Trio-Based Likelihood and...

Highly polymorphic markers like microsatellites are extensively utilized in genomic studies to analyze and infer genealogical relationships among individuals in a population. Traditional methods to identify the most likely parent among the potential known candidates rely on a single hypothetical distribution derived from population parameters. Indeed, these methods often make simplifying assumptions, such as a homogeneous genetic structure, consistent typing error rates across all genomic loci, and even random allele substitutions based on allele frequencies, which are frequently violated in practical applications. In this study, we introduce an enhanced likelihood-based approach, called the "Pairwise" algorithm, which builds on the widely used CERVUS method by calculating a trio-specific significance criterion for each father-mother-offspring combination using forward and backward simulations. Our method also accounts for the variable typing errors across genomic loci to enhance the accuracy of paternity analysis. Our findings showed that employing the Pairwise algorithm increases the power of paternity assignments by reducing the number of falsely assigned parents. Furthermore, adjusting likelihood equations to accommodate variable typing errors significantly improves the accuracy of paternity assignments. The developed approach represents a significant advancement in paternity analysis by addressing the limitations of traditional approaches. These improvements have the potential to significantly impact genealogical research and related fields, providing a more robust framework for analyzing complex genetic relationships in the context of parent assignment. Future research should focus on further refining this method and exploring its applications in diverse populations and genetic contexts.