An empirical approach to demographic inference with genomic data

Area of Science:

• Population Genetics
• Computational Biology
• Genomic Inference

Background:

• Traditional population genetic inference treats the population pedigree as a nuisance parameter.
• The historical network of genetic relationships is often averaged over in demographic modeling.
• An alternative perspective views the population pedigree as a complex, observable object.

Purpose of the Study:

• To explore an alternative approach to population genetic data inference by focusing on the population pedigree.
• To translate population genetic questions into calculations involving mutations on ancestral genomes.
• To develop robust interpretations for existing statistics and create new ones for demographic inference.

Main Methods:

• Translating population genetic inquiries into calculations using a Poisson process of mutations on ancestral genomes.
• Applying the method to re-interpret the f4 statistic for admixture identification.
• Designing a new statistic to quantify covariances in mean times to the most recent common ancestor.

Main Results:

• Provides a robust interpretation for the f4 statistic, enhancing admixture detection.
• Introduces a novel statistic for measuring covariances in mean coalescence times between sequence pairs.
• Generalizes population genetic statistics into interpretable sums over ancestral genomes.

Conclusions:

• Offers a method for describing demographic history without relying on simplified demographic models.
• Highlights the importance of the population pedigree, often overlooked in standard inference methods.
• Provides concrete, broadly interpretable insights into population genetic data and history.