From Nuisance to Signal: Leveraging Close Relatives in Biobank-Scale Demographic Inference.

Area of Science:

• Population genetics
• Bioinformatics
• Genomic data analysis

Background:

• Biobank-scale datasets are growing, increasing the prevalence of close relatives.
• Current population genetics practices often remove close relatives, treating them as nuisance parameters.
• The impact of removing relatives on demographic inference, particularly recent effective population size (Ne), is not well understood.

Purpose of the Study:

• To evaluate the consequences of including or excluding close relatives for recent effective population size (Ne) inference.
• To benchmark two widely-used methods (IBDNe and HapNe-IBD) for Ne estimation using identity-by-descent (IBD) segments.
• To investigate the effects of different relative sampling schemes on Ne estimates.

Main Methods:

• Benchmarking IBDNe and HapNe-IBD with simulated demographic histories and relative sampling schemes.
• Developing an open-source IBD simulation pipeline using msprime for realistic IBD segment generation.
• Analyzing the impact of removing relatives of varying degrees on Ne estimates and identifying sources of bias.

Main Results:

• Retaining all relatives in randomly ascertained samples yields the least biased Ne estimates.
• Removing even second-degree relatives inflates recent Ne and creates "ripple" artifacts, biasing estimates up to ten generations.
• Deliberately oversampling close relatives causes severe downward bias in recent Ne.

Conclusions:

• The practice of removing close relatives can lead to significant biases in recent effective population size inference.
• Retaining close relatives is generally the best practice for identity-by-descent-based Ne estimation in the biobank era.
• Guidelines for IBD simulation schemes incorporating pedigrees are provided to aid future research.