carmon - Copula Aided Reconstruction of Multi-Omics Networks

Carmon uses copula-based approaches to reconstruct multi-omics networks. Copulas allow to maximize the use of the information contained in non-normalized raw omics data, creating a mirror image in the normal realm for each variable of every omics layer. Carmon then uses these “mirror” normal variables to reconstruct the multi-omics network.

Installation

You can install carmon from Bioconductor with:

if (!require("BiocManager", quietly = TRUE))
    install.packages("BiocManager")

BiocManager::install("carmon")

Alternatively, you can install the development version from GitHub. First, make sure to install devtools with:

if (!require("devtools")) {
  install.packages("devtools")
}

You can then install the development version with:

devtools::install_github("DrQuestion/carmon")

Usage

Here is an example of the typical usage of carmon to reconstruct a multi-omics network. For a more customized use of the package, please see vignette("carmon"). The main function of the package is carmon(), a single wrapping function encompassing all the modules of the package. When the input data are properly formatted, carmon() only needs them as a mandatory argument. They should be preferably provided in the form of a named R list, with each element being the data set of an omics layer, named in the R list after the omics type it contains. The available omics types for which carmon has been specifically tailored, together with their accepted synonyms, are shown by the function which_omics(). The package expects every omics layer to be perfectly source-matched. This means that every variable of every omics layer must have been measured for the same individual/sample/observation/replicate. The package provides a few examples of multi-omics data sets formatted according to carmon’s standard. In this example we use the small data set already provided by the package, multi_omics_small.

# Attaching the package
library(carmon)

# Loading the carmon-formatted multi-omics data set
data(multi_omics_small)

# Displaying the structure of the named R list
str(multi_omics_small)

You can see that the data set has two layers, a transcriptomic layer named rnaseq in the list, and a metabolomic layer named metabolomics. This shows also that both layers have been measured over (the same) 30 observations, and how of they measure, respectively, the transcript counts of 14 genes and the concentrations of 5 metabolites. When the input is properly formatted, it is possible to just run the function carmon() on it, relying on its default behaviour.

carmon_obj <- carmon(multi_omics_small)

This single call covers all functional modules of the package. To see an example of how to properly format your input data, an explanation of each functional module of the package that acts under the hood of the main function carmon(), and examples of more customized behaviours, please see vignette("carmon").

Among the functional modules, one performs a centrality analysis based on a consensus of multiple centrality measures to highlight key nodes of the network. These nodes are candidates that could have an important function in the biological process you are studying. The report generated by the network analysis can be accessed via:

centrality_report(carmon_obj)

The report table contains the nodes that resulted to be central in the analysis, according to four different measures of centrality: degree (d), betwenness (b), closeness (c), and eigenvector (e) centrality. Names that result central according to more than one measure are reported on top of the table, and the last column contains the first letter(s) of the measure(s) according to which they resulted to be central. In our interpretation, the more measures label a node as central, the more important this node is in the network.

Contributing

The carmon package is built to grow. In particular, we intend to expand it to new omics types and relative marginal distributions. If you wish for your favourite omics layer to be included, please feel free to issue a feature request on the , or to directly contribute following the instructions in the page.

Code of Conduct

Please note that the carmon project is released with a Contributor Code of Conduct. By contributing to this project, you agree to abide by its terms.