README.md

In MSeidelFed/RandodiStats_package: Random distributions and their univariate and multivariate statistics

R package RandodiStats

Random distributions and their statistics

First step

It is necessary to install Rtools before starting. Make sure to install the appropriate version depending on your R version and add the Rtools directory to the system PATH (instructions for this can be found here)

or use the following code inside R studio.

installr::install.Rtools()

Installation

Install via "devtools"

library(devtools)

devtools::install_github("MSeidelFed/RandodiStats_package")

library(RandoDiStats)

Documentation of usage

?RandoDiStats::distribution_test_mat()

?RandoDiStats::testing_distributions()

?RandoDiStats::Variables2Shapes()

?RandoDiStats::plotting_distributions()

?RandoDiStats::ClustPlus()

?RandoDiStats::KmeansPlus()

?RandoDiStats::PlusPCA()

Needed objects

All our functions take as input a matrix with the same format, to check the format of the input matrix use the "distribution_test_mat()" function:

test_mat <- distribution_test_mat()

Dependencies

library(fitdistrplus)
library(raster)
library(lawstat)
library(matrixStats)
library(tidyr)
library(dendextend)
library(pvclust)
library(ComplexHeatmap)
library(circlize)
library(ggplot2)
library(reshape2)
library(ggfortify)
library(plotrix)
library(cluster)
library(multcompView)
library(arrangements)
library(pcaMethods)
library(viridis)

Examples

### this is how the input matrix should look like, metabolites in columns, treatments in rows (wide data)
test_mat <- distribution_test_mat()


### plot the distributions of your metabolites against common distributions
plotting_distributions(test_mat = distribution_test_mat2())


### test the distributions of your metabolites (this function is implemented into the Omics test to select an appropriate family for regression)
testing_distributions()

### obtain the geometrical shapes of your metabolite distributions (this is already implemented into the plotting and Omics testing functions)
Variables2Shapes()

### this needs to be a factor with the same size as rows in your columns and is meant to cover the treatment/s
Factor1_eg <- as.factor(c(rep("RED", 200), rep("GREEN", 200), rep("BLACK", 200),
                          rep("WHITE", 200), rep("YELLOW", 200)))

test_OUS <- OmicsUnivariateStats(Factor1 = Factor1_eg)

### run in the console, the function is interactive.

test_CP <- ClustPlus()

### k-means

test_Km <- KmeansPlus(DataDir = distribution_test_mat(nrow_x = 20, n_random_distributions = 10), n_boot = 1)

### PCA

PlusPCA(Km = KmeansPlus())

The recommended transformations are meant to cope with the intrinsic limitations of the different GLM families used for parametrization and do not alter the variance component of response variables, thus significances remain intact and reflect the biology of the treatments.

MSeidelFed/RandodiStats_package documentation built on July 31, 2022, 3 a.m.