Welcome to MARIPoSA’s documentation!

A platform for Manageable And Reproducible Integrated Pose Segmentation Analysis (MARIPoSA). See below for instructions on a quick start, or navigate through the tutorials and function descriptions for more detailed descriptions of the software functionality.

Quick Start

With GUI:

First, install the conda environment:

conda env create -f MARIPoSA/conda/environment.yaml

Activate the newly installed conda environment:

conda activate mariposa

Run the GUI:

python MARIPoSA/main.py

In the GUI:

1. Click “New project”

2. Enter project metadata.

   • NOTE: For “Path to data directory” - if you datatype is B-SOiD or Keypoints-MoSeq, you should enter the path to a folder containing all the label .csv files with their original names; if your datatype is VAME you should enter the path to the “results” folder ouptut by VAME which contains nested subdirectories of a structure like “video1/VAME/hmm-15/15_km_label_video1.npy”

3. Start your analysis!

With Python package:

Activate the newly installed conda environment

conda activate mariposa

Check out the demo scripts for a full analysis and plotting pipeline!

Contents:

• Demo 1 (PS/PE): Project Creation
  • 1.0 Import and Setup
  • 1.1: Creating a pose segmentation project
  • 1.2 Creating a pose estimation project
• Demo 2 (PS): Visualization of Pose Module Usage
  • 2.1: Loading data and simply visualizing usage (no subgroups or time binning)
  • 2.2: Loading data and visualizing usage by subgroups
• Demo 3 (PS): Embedding of Pose Segmentation Data
  • 3.1: Embedding with PCA
  • 3.2: Embedding with LDA
• Demo 4 (PS): Regression and Classification of Group Conditions in Pose Segmentation Data
  • 4.1: Classification
  • 4.2: Regression
• Demo 5 (PS): Comparing Pose Modules to Manual Scoring Data
  • 5.1: Remapping pose modules to grouped classes
  • 5.2: Incorporating manual scoring data from BORIS to remap modules
• Demo 6 (PS): Simulating Pose Segmentation Data
  • 6.1: Simulating time-series categorical pose segmentation data
  • 6.2: Simulating module usage
• Demo 7 (PE): Analyzing Keypoint Movement and Kinematics in Pose Estimation Data
  • 7.1: Measuring and plotting keypoint travel in animals
  • 7.2 Using keypoint travel in animals for embedding and classification
  • 7.3: Measuring and using ego-centered kinematics from pose estimation data
• Demo 8 (PE): Analyzing Keypoint Displacement and Action Units in human facial pose data from OpenFace
  • 8.0: Getting familiar with OpenFace data
  • 8.1 Measuring and Analyzing Keypoint Travel in OpenFace Data
  • 8.2: Measuring and using ego-centered facial kinematics from OpenFace data
  • 8.3 Working with Action Units
• Metadata
  • create_PE_project()
  • create_PS_project()
  • load_project()
  • save_edited_project()
• Analyze
  • BORIS_to_pose()
  • classify()
  • combine_pose_modules()
  • ego_center()
  • embed()
  • get_action_units()
  • get_distance()
  • get_keypoint_kinematics()
  • get_keypoint_travel()
  • get_module_labels()
  • get_module_transitions()
  • get_module_usage()
  • loocv()
  • loocv_regression()
  • make_remappings_from_BORIS()
  • regress()
• Plot
  • module_usage_sandplot()
  • network_plot()
  • plot_action_units()
  • plot_distance_box()
  • plot_distance_matrix()
  • plot_distance_results()
  • plot_embeddings()
  • plot_keypoint_kinematics()
  • plot_keypoint_travel()
  • plot_module_usage()
• Simulate
  • generate_sequence()
  • generate_usage()
  • generate_usage_labeled()