NPflow
Bayesian Nonparametrics for Automatic Gating of Flow-Cytometry Data
Dirichlet process mixture of multivariate normal, skew normal or skew t-distributions
modeling oriented towards flow-cytometry data preprocessing applications. Method is
detailed in: Hejblum, Alkhassimn, Gottardo, Caron & Thiebaut (2019)
Overview
NPflow is a package for performing Bayesian estimation of Dirichlet process mixtures of multivariate skew t-distributions. It is especially oriented towards flow-cytometry data preprocessing applications, but can be useful for numerous other applications.
The main function of the package is DPMpost().
The method implemented in this package is detailed in the following article:
Installation
The easiest way to get NPflow is to install it from CRAN:
install.packages("NPflow")
Or to get the development version from GitHub:
devtools::install_github("borishejblum/NPflow")
-- Boris Hejblum
News
News about the NPflow R package
Main changes in Version 0.13.1 (2017-08-02) --- this is only a minor release:
- updated documentation and compiling standards
Main changes in Version 0.13.0 (2017-04-10):
- use of
itertoolspackage for parallel computations - use of log-scale probabilities in slice-samplers for improved precision in computation
- bug fix: parallel settings did not work in DPMpost (thanks to Mike Jiang for noticing this)
- updated documentation
Main changes in Version 0.12.0 (2017-04-04) --- this is only a minor release:
- registration of compiled functions
- updated help and citation info
Main changes in Version 0.10.0 (2016-05-02):
- optimization of the log-posterior computation for normal and skew normal distributions
- improvement of the parallel sampler for normal distributions
- added arguments in
DPMGibs...functions to allow more informative priors if needed - bug fix: log of integers disambiguated in C++ function (compilation error on Solaris)