knitr::opts_chunk$set( collapse = TRUE, comment = "#>" )
The waddR
package offers statistical tests based on the 2-Wasserstein distance for detecting and characterizing
differences between two distributions given in the form of samples. Functions for calculating
the 2-Wasserstein distance and testing for differential distributions are provided, as well as a specifically
tailored test for differential expression in single-cell RNA sequencing data.
waddR
provides tools to address the following tasks, each described in a separate vignette:
Two-sample tests to check for differences between two distributions,
Detection of differential gene expression distributions in single-cell RNA sequencing (scRNAseq) data.
These are bundled into one package, because they are internally dependent: The procedure for detecting differential distributions in scRNAseq data is an adaptation of the general two-sample test, which itself uses the 2-Wasserstein distance to compare two distributions.
The 2-Wasserstein distance is a metric to describe the distance between two
distributions, representing e.g. two diferent conditions $A$ and $B$. The waddR
package
specifically considers the squared 2-Wasserstein distance which can be decomposed into location, size, and shape terms, thus providing
a characterization of potential differences.
The waddR
package offers three functions to calculate the (squared)
2-Wasserstein distance, which are implemented in C++ and exported to R with Rcpp for faster
computation.
The function wasserstein_metric
is a Cpp reimplementation of the
wasserstein1d
function from the R package transport
.
The functions squared_wass_approx
and squared_wass_decomp
compute
approximations of the squared 2-Wasserstein distance, with squared_wass_decomp
also returning the decomposition terms for location, size, and shape.
See ?wasserstein_metric
, ?squared_wass_aprox
, and ?squared_wass_decomp
for more details.
The waddR
package provides two testing procedures using the 2-Wasserstein distance
to test whether two distributions $F_A$ and $F_B$ given in the form of samples are
different by testing the null hypothesis $H_0: F_A = F_B$ against the
alternative hypothesis $H_1: F_A != F_B$.
The first, semi-parametric (SP), procedure uses a permutation-based test combined with a generalized Pareto distribution approximation to estimate small p-values accurately.
The second procedure uses a test based on asymptotic theory (ASY) which is valid only if the samples can be assumed to come from continuous distributions.
See ?wasserstein.test
for more details.
The waddR
package provides an adaptation of the semi-parametric testing procedure based on the
2-Wasserstein distance which is specifically tailored to identify differential distributions in scRNAseq
data. In particular, a two-stage (TS) approach is implemented that takes account of the specific
nature of scRNAseq data by separately testing for differential proportions of zero gene expression
(using a logistic regression model) and differences in non-zero gene expression (using the semiparametric
2-Wasserstein distance-based test) between two conditions.
See ?wasserstein.sc
and ?testZeroes
for more details.
To install waddR
from Bioconductor, use BiocManager
with the following commands:
if (!requireNamespace("BiocManager")) install.packages("BiocManager") BiocManager::install("waddR")
Using BiocManager
, the package can also be installed from GitHub directly:
BiocManager::install("goncalves-lab/waddR")
The package waddR
can then be used in R:
library("waddR")
sessionInfo()
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.