An R Package for Nonparametric Inference on Dynamic Populations with Infinitely Many Types

Area of Science:

• Population genetics
• Stochastic modeling
• Computational biology

Background:

• Fleming-Viot diffusions model population dynamics and allele frequency changes.
• Bayesian nonparametric inference is increasingly applied to these models with finite population samples.
• Existing software is limited to simpler Wright-Fisher models.

Purpose of the Study:

• To develop and provide software for Bayesian inference in general Fleming-Viot diffusion models.
• To address computational challenges in analyzing complex population genetic data.
• To enable efficient filtering and smoothing of allele frequencies over time.

Main Methods:

• Implementation of an R package (FVDDPpkg) for Fleming-Viot diffusions.
• Development of algorithms for approximating filtering and smoothing distributions.
• Introduction of Monte Carlo methods to reduce computational cost.

Main Results:

• The FVDDPpkg package provides efficient approximations for filtering and smoothing distributions.
• The software handles general Fleming-Viot diffusion models with finite samples.
• Computational efficiency is improved through Monte Carlo approximations.

Conclusions:

• The FVDDPpkg package democratizes advanced population genetic analysis.
• This software facilitates research in evolutionary biology and population dynamics.
• It overcomes significant computational hurdles in Bayesian inference for complex models.