Networks in R

Kelsey Andersen, Robin Choudhury, Karen Garrett

What is a network? Some key terms

• Nodes

• Edges

• Dyads – two nodes

• Triads – three nodes

Node

Link

Also known

as a vert e x

Also known as

an edge

Applications of Networks in Plant

Pathology

• Social networks – networks of connection between individuals

(researchers, producers, land managers)

• Links in this case could represent exchange of goods, information, seed, etc.

• Trade networks – networks of plant movement locally and globally

pose major phytosanitary risks

• Networks of association of microbes – viromes, microbiomes, etc.

• Gene co-expression networks

Networks 101: A network is basically

an adjacency matrix

N1 N2 N3

From

Node1 0 1 1

Node2 0 0 0

Node3 1 0 0

Can you draw what this network would look

like?

N1 N2 N3

From

N1 0 0 0

N2 1 0 1

N3 0 0 0

Directed or undirected network?

N1 N2 N3

From

N1 0 1 0

N2 0 0 1

N3 0 1 0

N1 N2 N3

From

N1 0 1 0

N2 1 0 1

N3 0 1 0

Symmetric!

Examples of general questions to ask

with networks

• What nodes and links have particularly important roles in the

network?

• Are there clusters of nodes?

• How “complex” is the network?

• What would strengthen or weaken the network?

Today we will do most of our work in igraph

• igraph reference manual

• http://igraph.org/c/doc/igraph-introduction.html

• Katya Ognyanova has prepared some user friendly introductions

• Network visualization: http://kateto.net/network-visualization

• igraph in R: http://kateto.net/networks-r-igraph

• There are other options, such as ggnetwork, but there is less

documentation available for these packages

Creating networks with igraph

• We can start with a simple adjacency matrix:

Specify number of columns

and rows

Quickly visualize our network using plot()

Network Aesthetics

Some important igraph plotting parameters

Nodes

vertex.color: color of node

vertex.frame.color: border color of node

vertex.shape: shape of node

vertex.label.family: font type

vertex.label.dist: distance of label from node

vertex.label.cex: font size of label

Edges

edge.color: color of edge

edge.width: size of edge

edge.arrow.size: arrow size (for directed networks)

edge.curved: curved edge

Adjusting labels

Curving and coloring edges

Changing shapes of nodes and edges

Other ways to adjust network aesthetics

V = Vertex

E = Edge

Network data types: nodelist and edgelist

• Adjacency matrices can get large quickly

• Another way to read in data is to use an edgelist and a nodelist

Here is our nodelist for a producer network:

First column in “node ID”

Network data types: nodelist and edgelist

“node ID” to “node ID”

Create an igraph object

Edgelist

Nodelist

**All Node ID’s must be

present in the nodelist

Check your

network

Associating colors with node attributes

Color node and

node frame

based on

gender

Adjusting and saving layouts

(Kamada-Kawai layout algorithm)

Sizing nodes based on

attributes

Scale by 2

Size links based on link attributes

Size links based on the

number of times producers

communicated

Fix layout with one of our

previous layouts

Adding a legend

Saving an igraph object