Detail: Display Subgraphs

Display Subgraphs dialog for extracting subgraphs from a larger graph.

Purpose

The Display Subgraphs tool is designed to help you explore large, complicated graphs by extracting and visualizing subgraphs of interest. Instead of working with the entire graph at once, you can:

• Select a subset of variables,

• Choose a subgraph type (e.g., neighbors, ancestors, paths, Markov blankets),

• Click Graph It! to open a new Graph window containing only the relevant part of the graph.

This is invaluable for understanding local neighborhoods, causal families, and path structures in large graphs.

Basic workflow

1. Open Display Subgraphs

   From a Graph window, choose the Display Subgraphs tool (for example, via the Graph menu).

2. Select variables

   • The dialog shows two lists:

     • Not selected – variables currently excluded from the subgraph,

     • Selected – variables currently included.

   • Use the > and < buttons to move variables between lists.

   • Sort Variables orders the variable names alphabetically.

   • Text Input… lets you specify a list of variables by name.

3. Choose a subgraph type

   • At the bottom, use the graph type drop-down to choose what kind of subgraph to construct.

   • Depending on the type, all nodes and edges that satisfy the requested relationship (adjacency, ancestry, path relation, degree condition, and so on) are included in the subgraph.

4. (Optional) Degree filter

   • For degree-based options you can specify:

     • Whether the degree should equal, be at most, or at least a given number,

     • The numerical threshold in the text field.

5. Graph it

   • Click Graph It! to create and display the subgraph in a new Graph window.

   • The original graph is left unchanged; you can open multiple subgraph windows for comparison.

   • Click Done to close the dialog when you are finished.

Subgraph types

The graph-type drop-down offers the following subgraph types:

Neighborhood-based selections

• Subgraph
  Shows the induced subgraph on the selected variables: all selected nodes and all edges between them.

• Adjacents
  For the selected variables, includes each selected node plus all of its neighbors (adjacent nodes) and the edges among them.

• Adjacents of Adjacents
  Extends Adjacents one step further: includes selected nodes, their neighbors, and the neighbors of those neighbors.

• Adjacents of Adjacents of Adjacents
  Extends the neighborhood out to three adjacency steps from the selected nodes.

Family / causal-relation selections

• Parents
  Selected nodes plus all of their parents and the edges between them.

• Children
  Selected nodes plus all of their children and the edges between them.

• Ancestors
  Selected nodes plus all of their ancestors (nodes with a directed path into them) and all connecting edges.

• Descendants
  Selected nodes plus all of their descendants (nodes reachable by directed paths out of them) and the relevant edges.

• Markov Blankets
  For each selected node, includes its Markov blanket (typically its parents, its children, and the parents of its children), along with all edges among these nodes.

Path-based selections

These focus on paths and treks involving the selected variables.

• Treks
  Subgraph consisting of treks (paths without colliders) involving the selected nodes.

• Trek Edges
  Only the edges that lie on those treks, together with their endpoints.

• Paths
  Subgraph containing paths involving the selected nodes.

• Path Edges
  Only the edges that lie on such paths, together with their endpoints.

• Directed Paths
  Subgraph consisting of directed paths involving the selected variables.

• Directed Path Edges
  Only the edges that lie on those directed paths, together with their endpoints.

• Y Structures
  Highlights Y-structures (collider patterns of the form X → Z ← Y with no edge between X and Y) that involve the selected variables.

• Pag Y Structures
  The corresponding Y-structure patterns in PAGs, using PAG edge types.

Degree-based selections

These options select nodes based on their degree properties.

• Indegree
  Nodes whose indegree (number of incoming arrows) satisfies the chosen condition (equal / at most / at least some integer).

• Out Degree
  Nodes whose outdegree (number of outgoing arrows) satisfies the chosen condition.

• Degree
  Nodes whose total degree (number of incident edges, regardless of direction) satisfies the chosen condition.

In each case, you can specify the condition using the Degree controls at the bottom of the dialog: Equals, At Most, or At Least, together with a numeric threshold.

Summary

The Display Subgraphs tool lets you:

• Focus on local neighborhoods (adjacents and relatives),

• Examine causal families (parents, children, ancestors, descendants, Markov blankets),

• Study path structures (paths, treks, directed paths, Y-structures),

• Filter by degree properties.

It is especially powerful for large graphs, where directly seeing all of these structures in the full graph would be difficult or impossible.