An approximate full-likelihood method for inferring selection and allele frequency trajectories from DNA sequence...

Area of Science:

• Evolutionary Biology
• Population Genetics
• Bioinformatics

Background:

• Current methods for detecting natural selection from DNA sequence data often rely on summary statistics or composite likelihoods, limiting their use of available information.
• These limitations hinder accurate detection and characterization of selection events.

Purpose of the Study:

• To introduce a novel importance sampling approach, CLUES (Coordinated Likelihood Using Importance Sampling), for approximating the full likelihood function of the selection coefficient.
• To enhance the detection of natural selection, estimation of selection coefficients, and inference of allele frequency trajectories from DNA sequence data.

Main Methods:

• CLUES treats the ancestral recombination graph (ARG) as a latent variable, integrating it out using Markov Chain Monte Carlo (MCMC) methods.
• The method approximates the full likelihood function for the selection coefficient, enabling a more comprehensive analysis of DNA sequence data.

Main Results:

• Extensive simulations demonstrate that CLUES uniformly improves power for detecting selection compared to popular methods like nSL and SDS.
• CLUES provides reliable inferences of allele frequency trajectories under various conditions and can detect recent changes in selection strength.
• Application to human populations revealed insights into lactase persistence (MCM6), EDAR SNP age in Han Chinese, and selection on pigmentation genes (ASIP, KITLG, TYR, OCA2/HERC2).

Conclusions:

• CLUES represents a significant advancement in detecting natural selection and inferring evolutionary histories from DNA sequence data.
• The method's ability to utilize the full information in DNA sequences and its improved performance over existing techniques make it a valuable tool for population geneticists.
• Analysis of specific human genes highlights the method's utility in uncovering detailed patterns of recent and ancient selection.