Pedigree reconstruction from SNP data: parentage assignment, sibship clustering and beyond

Area of Science:

• Genetics
• Bioinformatics
• Computational Biology

Background:

• High-throughput genotyping generates vast amounts of single nucleotide polymorphism (SNP) data.
• Existing tools struggle to effectively construct multigenerational pedigrees from dense SNP data.

Purpose of the Study:

• Introduce the R package sequoia for accurate multigenerational pedigree reconstruction.
• Enable the assignment of parentage, half-siblings, and grandparents from SNP data.

Main Methods:

• Utilizes a heuristic hill-climbing algorithm to explore relationship likelihoods.
• Calculates likelihoods for first-, second-, and third-degree relationships, including unrelated individuals.
• Incorporates genotyping error rates and missing data in performance evaluations.

Main Results:

• Parentage assignment is highly accurate with ~100 independent SNPs (error rate <0.1%).
• Full pedigree reconstruction achieves low error rates (<0.3%) and high assignment rates (>99%) with ~200 independent SNPs.
• Inferred pedigrees show strong correlation with genomic relatedness in empirical datasets.

Conclusions:

• The sequoia R package provides a fast and accurate method for multigenerational pedigree reconstruction using SNP data.
• It effectively handles complex pedigrees with missing parents, inbreeding, and unknown birth years.
• The tool has broad applications in population genetics, evolutionary studies, and conservation.