tidygraph R Package Documentation

tidygraph provides a tidy API for graph/network manipulation by treating a graph as two relational tables: one for node data and one for edge data.

It wraps igraph functionality into a dplyr-compatible interface. You use the activate() function to switch the context of your verbs between nodes and edges. Once activated, dplyr verbs like mutate() can be used to add properties to nodes or edges using graph algorithms that are context-aware.

library(tidygraph)

play_gnp(10, 0.5) %>%
  activate(nodes) %>%
  mutate(degree = centrality_degree()) %>%
  activate(edges) %>%
  mutate(centrality = centrality_edge_betweenness()) %>%
  arrange(centrality)

The morph() and unmorph() verbs allow you to temporarily change the representation of a graph (e.g., contracting nodes or working on a linegraph), perform manipulations on that representation, and then automatically merge those changes back into the original graph structure.

If you want to keep the transformed representation as a permanent tbl_graph instead of merging it back, use the crystallise() verb.

tidygraph wraps igraph algorithms into functions that are designed to be used inside dplyr verbs. These functions are context-aware, meaning they automatically detect the graph and the relevant node/edge IDs from the current tidygraph object, resulting in cleaner code.

Example usage:

# Inside a mutate call, the algorithm knows which nodes/edges to operate on
graph %>% 
  activate(nodes) %>% 
  mutate(deg = centrality_degree())

You can install the stable version of tidygraph from CRAN using install.packages(). For the development version from GitHub, it is recommended to use the pak package.

To install the development version:

pak::pak('thomasp85/tidygraph')

The activate() function is used to define whether subsequent operations are performed on node data or edge data. This is a prerequisite for using dplyr verbs like mutate() or filter() on specific parts of the graph.

Common arguments for activate():

• nodes
• edges

# Switch to node context
graph %>% activate(nodes) %>% mutate(new_col = ...)

# Switch to edge context
graph %>% activate(edges) %>% mutate(new_col = ...)

tidygraph provides several verbs to modify the structure of a graph beyond simple attribute manipulation:

• bind_edges(): Expand the graph structure by adding edges.
• bind_nodes(): Expand the graph structure by adding nodes.
• bind_graphs(): Combine multiple graphs.
• graph_join(): Merge two graphs based on a shared node identifier.
• reroute(): Change the terminal nodes of the edges in the graph.

The as_tbl_graph() function converts various relational and network objects into a tidygraph object (tbl_graph). Supported types include:

• network
• phylo
• dendrogram
• data.tree
• graph (igraph objects)