Implement Fleming-Viot-Dependent Dirichlet Processes

A Bayesian Nonparametric model for the study of time-evolving frequencies, which has become renowned in the study of population genetics. The model consists of a Hidden Markov Model (HMM) in which the latent signal is a distribution-valued stochastic process that takes the form of a finite mixture of Dirichlet Processes, indexed by vectors that count how many times each value is observed in the population. The package implements methodologies presented in Ascolani, Lijoi and Ruggiero (2021) and Ascolani, Lijoi and Ruggiero (2023) that make it possible to study the process at the time of data collection or to predict its evolution in future or in the past. The package itself is described in Ascolani, Damato and Ruggiero (2024) .

FVDDPpkg

R implementation of Fleming-Viot-dependent Dirchlet Processes

:computer: The package contains the first implementation of Fleming-Viot dependent Dirichlet processes, a model of Bayesian Nonparametric statistics used in populuation genetics.
The model consists of a Hidden Markov Model (HMM) in which the latent signal is a distribution-valued stochastic process that takes the form of a finite mixture of Dirichlet processes, indexed by vectors that count how many times each value is observed in the population.

In particular, the package allows to:

• implement a Dirichlet process specifying
  • the baseline prior distribution $P_0$, which is described by its p.m.f (if it is discrete) or by its p.d.f. (it is is absolutely continuous), and by a function to sample from.
  • the intensity $\theta > 0$ associated to $P_0$.
• update the structure of the model with data, modifying the latent signal.
• propagate the latent signal in the past or in the future for any time, given the current state.
• smooth the latent signal given observations from the past, the present and the future.
• sample from the process.
• exploit the predictive structure to make inference.

For clarifications and examples of use, please refer to the vignette in the package documentation or on GitHub.

The package is described in Ascolani, Damato and Ruggiero (2024). For a rigorous theoretical approach, read Ascolani et al. (2021) and Ascolani et al. (2023)

Installation

:space_invader: The package has been released on CRAN!

To download the latest version, just run

install.packages("FVDDPpkg", dependencies = True)

:cat: To install from GitHub, make sure you have the devtools package installed.

install.packages("devtools")

and proceed to download using

devtools::install_github("StefanoDamato/FVDDPpkg", build_vignettes = TRUE)

Citation

:book: If you use the package, please cite Ascolani, Damato and Ruggiero (2024), which presents the package and the algorithms it implements:

@article{AscolaniDamatoRuggiero2024,
author = {Filippo Ascolani and Stefano Damato and Matteo Ruggiero},
title = {{An R Package for Nonparametric Inference on Dynamic Populations with Infinitely Many Types}},
volume = {31},
journal = {Journal of Computational Biology},
number = {12},
pages = {1305 -- 1311},
year = {2024},
doi = {10.1089/cmb.2024.0600},
URL = {https://doi.org/10.1089/cmb.2024.0600}
}

Contacts

:e-mail: For questions, contributions or bug reports contact me at [email protected] or use the issues section on the GitHub repository.