MultiWaver 2.0: modeling discrete and continuous gene flow to reconstruct complex population admixtures

Area of Science:

• Population Genetics
• Computational Biology
• Human Evolutionary Studies

Background:

• Previous MultiWaver software versions were limited to discrete admixture models.
• Inferring complex population admixture histories requires flexible computational frameworks.

Purpose of the Study:

• To introduce MultiWaver 2.0, a novel software for inferring population admixture history.
• To implement a flexible framework capable of handling both discrete and continuous admixture models.
• To enable automatic selection of the optimal admixture model and estimation of its parameters.

Main Methods:

• Developed MultiWaver 2.0 with a flexible framework for discrete and continuous admixture models.
• Implemented an automated model selection based on chromosomal ancestral track length distribution.
• Utilized Likelihood Ratio Test (LRT) for discrete model selection and Expectation-Maximization (EM) algorithm for parameter estimation.
• Employed Bayesian Information Criterion (BIC) for comparing discrete and continuous models.
• Incorporated bootstrapping for model support and confidence intervals (CI) for admixture time estimations.

Main Results:

• MultiWaver 2.0 successfully infers multiple-wave admixture histories under both discrete and continuous models.
• The software automatically selects the optimal admixture model and estimates parameters accurately.
• Simulation studies confirmed the reliability and effectiveness of the MultiWaver 2.0 method.
• The program demonstrated robust performance on real-world admixed population datasets (African Americans, Uyghurs, Hazaras).

Conclusions:

• MultiWaver 2.0 provides a significant advancement in inferring complex population admixture histories.
• The software's automated model selection and parameter estimation enhance the analysis of population genetics data.
• MultiWaver 2.0 is a valuable tool for studying the demographic past of admixed populations.