A method for genome-wide genealogy estimation for thousands of samples

Area of Science:

• Population Genetics
• Computational Biology
• Human Evolution

Background:

• Genome-wide genealogies are crucial for evolutionary analysis but computationally challenging for large datasets.
• Previous methods lacked scalability for analyzing thousands of individuals.

Purpose of the Study:

• To develop a scalable computational method for inferring genome-wide genealogies.
• To analyze human population history and natural selection using large-scale genetic data.

Main Methods:

• Developed 'Relate,' a novel computational method for estimating genealogical relationships, branch lengths, and population sizes.
• Applied Relate to over 10,000 haplotypes from the 1000 Genomes Project.
• Simultaneously estimated demographic parameters and accounted for potential data errors.

Main Results:

• Generated joint genealogical histories for 26 human populations.
• Identified highly diverged lineages, more frequent in Africa, reflecting Neanderthal/Denisovan introgression outside Africa and unique African events.
• Detected multiple genomic regions under strong positive selection and directional selection on traits like hair color, BMI, and blood pressure.

Conclusions:

• The Relate method offers unprecedented power for inferring population histories and selection.
• Ancient introgression and continent-specific events significantly shaped human genetic diversity.
• Human traits show evidence of varying directional selection across different populations.