A NOVEL SPECTRAL METHOD FOR INFERRING GENERAL DIPLOID SELECTION FROM TIME SERIES GENETIC DATA

Area of Science:

• Population Genetics
• Evolutionary Biology
• Bioinformatics

Background:

• Analyzing time series genetic variation data is crucial for identifying genomic regions under selection.
• Estimating fitness parameters from temporal DNA data is computationally challenging, especially for non-neutral models.
• Existing methods often struggle with numerical approximations and are limited to specific selection models.

Purpose of the Study:

• To develop a novel, efficient, and flexible algorithm for analyzing population allele frequency dynamics.
• To overcome limitations of existing methods in handling complex selection models and numerical integration.
• To apply the new method to simulated and real ancient DNA data.

Main Methods:

• Developed a spectral algorithm to analytically integrate over possible allele frequency trajectories.
• The method handles general diploid models with arbitrary fitness parameters for heterozygotes and homozygotes.
• Applied the algorithm to simulated datasets and ancient DNA data concerning horse coat coloration loci.

Main Results:

• The spectral algorithm efficiently computes likelihoods for non-neutral models without extensive parameter space discretization.
• Demonstrated the method's utility on simulated data, confirming its accuracy and efficiency.
• Analysis of ancient horse DNA suggests a heterozygote-advantage form of balancing selection at coat coloration loci.

Conclusions:

• The novel spectral algorithm provides a significant advancement in analyzing temporal genetic variation data.
• This method offers greater flexibility in modeling selection compared to previous approaches.
• New insights into the evolutionary dynamics of horse coat color genes were uncovered, highlighting balancing selection.