Overview

This section provides step-by-step tutorials for using CoMPASS-Labyrinth to analyze behavioral data from the labyrinth navigation task.

Available Tutorials

00. DLC Grid Processing

Process video data with DeepLabCut, create spatial grids, and annotate trajectories with grid locations for labyrinth navigation analysis.

01. Create Project

Initialize a new CoMPASS-Labyrinth project, ingest DeepLabCut results, and preprocess combined session data for downstream analysis.

02. Task Performance Analysis

Analyze task performance metrics including spatial heatmaps, Shannon entropy, region usage, bout-level success rates, and deviation from optimal paths.

03. Simulated Agent Modelling

Compare animal navigation strategies to simulated agents using chi-square analysis, multi-agent comparisons, and exploration-exploitation modeling.

04. CoMPASS Level 1

Fit Hidden Markov Models to infer fine-grained motor states (surveillance vs. ambulation) from step length and turn angle distributions.

05. CoMPASS Level 1 Post-Analysis

Perform post-hoc analysis of Level 1 HMM results including bout-level state analysis, spatial mapping, and temporal dynamics visualization.

06. CoMPASS Level 2

Apply hierarchical modeling to integrate multiple behavioral and physiological data streams for multi-scale inference of cognitive states.

---

Prerequisites

• Python environment with CoMPASS-Labyrinth installed
• DeepLabCut pose estimation results (or raw videos for Tutorial 00)
• Project metadata file (Excel/CSV format)
• Basic familiarity with Jupyter notebooks

For installation instructions, see the Installation Guide.