In MetClassNet/MetClassNetR: Create and analize networks built from metabolomics Data

```{cat, include = FALSE} \tikzexternalize

```r
knitr::opts_chunk$set(echo = TRUE)

library(MetClassNetR)
library(tidyverse)
library(ggrepel)
library(plyr)

Networks' comparison

This script compares several aspects of a set of networks. We start by reading
the networks, using the readNet function, which gets as input the directory in which the networks are stored (netDir parameter). Each network should be an independent file, and several formats are accepted (e.g., CSV and GML). Please note that we assume that the networks will have equivalent node
names (i.e., they should all be subsets of a common set of nodes). Importantly,
you can define whether the networks are directed or not using the directed parameter.

NOTE. You can also generate a predefined set of 4 toy-networks with the makeToyNet function, giving as parameter the path of the folder where you want to store them.

# create toy networks
netDir <- tempdir()
makeToyNet(netDir)
directed <- FALSE

# read networks
networks <- readNet(netDir, directed = directed, format = "csv")

Now that we have the networks to analyze, we will proceed to calculate some basic statistics measurements and to make a plot for each of them. In order to do so, use the calculNetStats function, giving a parameter the networks you just read with readNet. To generate the plots, we will use the PrintStatsPlots function, which needs the results of the calculNetStats function as input parameter.

The statistics that will be calculated and displayed are the following:

• Density: Radio of number of edges and the number of possible edges (i.e., the number of edges of the fully connected graph).
• Diameter: Size of the shortest path between the furthest vertices, i.e., the length of the longest geodesic distance.
• Average path length: Mean of the shortest paths between each pair of vertices.
• Clustering coefficient: Number of triangles (i.e., node A connected to B and C, and node C connected to B) in the graph. It is a measure of the degree to which nodes in a graph tend to cluster together. In other words, the clustering coefficient tells you how well connected the neighborhood of every node is. It goes from 0 (i.e., no connections between the nodes in the neighborhood of each node) to 1 (i.e., the neighborhood of each node is fully connected).
• Degree: Number of edges connected to a node (i.e., its adjacent edges).
• Closeness: Reciprocal of the average shortest paths between a node and all the other nodes in a graph. It is a measure of the centrality of a node. When normalized, it allows comparing the closeness between graphs of different sizes. NOTE. The higher the closeness of a node, the more central that node is.
• Betweenness: Number of shortest paths going through a node or an edge. We calculate the betweenness of the nodes.

# calculate stats and generate plots
statistics <- calculNetStats(networks)

# print table with stats 
knitr::kable(statistics$netStats, caption = "General stats")

# print plots
printStatsPlots(statistics)

We will now calculate the overlap of the nodes and edges of all of our networks, using the CalculateOverlap function, which needs the networks to analyze as input parameter. The CalculateOverlap function the will also generate two upset plots, one to show the overlap of nodes, and the other for the edges. It is to note that the overlap of edges does not take direction into account and it will thus consider the networks as undirected.

Please note that the upset plots show an intersection of sets, but also a difference, e.g., the upset plot of the overlap of nodes between 2 networks, will show: The common nodes (i.e., intersection) of nodes in both networks The nodes present in the first network, but not in the second one (difference), and * The nodes present in the second network, but not in the first one.

# calculate overlap of nodes and edges
calculateOverlap(networks)

If you want to get the list of overlapping nodes, you can use the getOverlappingNodes function, giving as input the list of networks and the list of index of the networks you want to take into account. NOTE. If no list of index is provided, all the networks are considered for the calculus of the overlap.

# get list of overlapping nodes 
getOverlappingNodes(networks)

MetClassNet/MetClassNetR documentation built on June 30, 2023, 2:12 p.m.