In edlansiaux/muinther: Mutual Information Theory and Correlation Computation for Numerous Variables

knitr::include_graphics(system.file("logos","muinther_hex.png", package = "muinther"))
options(rmarkdown.html_vignette.check_title = FALSE)

muinther is a package to perform analysis concerning correlation and association between numerous variables (e.g., binary and/or quantitative variables).The package implements two distinct strategies to do that: 1) Pearson's correlation computation and 2) Shannon mutual information method. For both strategies function provide a contingence heatmap matrix with two values for each box : one (p-value of association/correlation test) on a numeric scale and the other one (association/correlation coefficient) on a color scale. This contingence heatmap matrix was built thanks to ggplot2 package.

The methods implemented in this package are described in detail in the following publications.

1. Edouard Lansiaux, Jean-Luc Caut, Joachim Forget et al. Assessing the efficiency of COVID-19 NPIs in France: a retrospective study using a novel methodology, 13 April 2021, PREPRINT (Version 1) available at Research Square [https://doi.org/10.21203/rs.3.rs-321360/v1]

Below, we provide a quick-start guide using a data set to illustrate the functionalities of the muinther package.

Quick start

A standard muinther analysis takes the following form, where docs_phenotype_file_1 represents a matrix or data.frame of (of dimension d x n for d observations and n studied variables).

Mutual information theory results (computed thanks to loop()), exported in the form of a entropy_outputs csv object, can be easily be examined using heatmap() after csv file transformation into dataframe/matrix (see below and the User's Guide for example).

library(muinther)
pearsontable(docs_phenotype_file_1)
loop(docs_phenotype_file_1,1,8)
entropy_outputs <- readr::read_csv('entropy_outputs.csv')
heatmap2(entropy_outputs)

edlansiaux/muinther documentation built on Dec. 24, 2024, 4:46 p.m.