Quickstartο
A five-minute walkthrough: load data β run a search β view your first causal graph.
This guide assumes the Tetrad GUI is already running. (If not see the Tetrad GUI setup page.)
1. Add a Data or Simulation Boxο
Option A β Load dataο
Add a Data box from the left toolbar.
Double-click it β Load Dataβ¦
Choose a CSV/TSV/whitespace file (or a covariance matrix).
Confirm variable types if prompted.
Your dataset now appears in the Data box.
Option B β Simulate dataο
Add a Simulation box.
Choose a generator (linear SEM, nonlinear SEM, etc.).
Enter number of variables and sample size.
Click Simulate.
A graph and dataset appear inside the Simulation box.
2. (Optional) Add Prior Knowledgeο
If you have background knowledge:
Add a Knowledge box.
Draw an arrow from the Data/Simulation box β to the Knowledge box.
Specify required edges, forbidden edges, or tiers.
You may skip this step for your first run.
3. Run a Causal Searchο
Add a Search box.
Draw an arrow from the Data box β to the Search box
(and from the Knowledge box if youβre using one).Double-click the Search box and choose an algorithm:
PC, FGES, BOSS β if no hidden confounders (DAG/CPDAG)
FCI, GFCI, BFCI β if hidden confounders are possible
Set parameters if needed.
Click Run Search and Generate Graph.
A completed Search box now contains the learned graph.
4. View and Edit the Graphο
To work with the graph directly:
Add a Graph box to the workbench.
Draw an arrow from the Search box β to the Graph box.
Double-click the Graph box to view the graph.
Here you can:
Add/remove/orient edges
Highlight colliders or paths
Check whether itβs a DAG/CPDAG/PAG
Inspect properties and adjustment sets
(You can preview the graph inside the Search box, but the Graph box is where full interaction happens.)
5. Save and Exportο
Save the session (File β Save Session, or Save Session As)
Export graphs (Text, JSON, other formats)
You now have a full causal workflow:
data β search β editable graph.