nhat-le / block-hmm-simulations

General information

Code tested on MAC OS 10.15.7, MATLAB 2020a, and Python 3.9

Instructions for running the analysis and simulation scripts

Basic setup

0. Install MATLAB, python, anaconda and jupyter notebook if these have not been installed
1. Clone the directory git clone https://github.com/nhat-le/switching-simulations
2. Add the folder block-hmm-simulations to the MATLAB path
3. Install the following Python packages: smartload: python -m pip install [package_name]
4. Install the following fork of the ssm package: https://github.com/nhat-le/ssm (Note: modified from original package https://github.com/lindermanlab/ssm)

Running blockHMM on synthetic data

• To generate and fit the synthetic data, run blockHMM_simulations.ipynb.

Raw data files will be saved in the file blockhmm_simulated.mat. Data for the K-selection will be saved in blockhmm_synthetic_K_validation.mat.

• Then run the MATLAB script blockhmm_synthetic.m to generate the figures.

Instructions for data analysis code for the paper

Code for producing the figures in the paper can be run by calling the scripts figX.m in MATLAB. Data required for running the code is available here

Some jupyter notebooks are provided for data pre-processing:

• session_average_parameters_per_animal.ipynb generates the summary parameter fits of session-averaged transition functions of animals (Fig. 1)

• switching_world_classifier.ipynb performs the forward simulations that generate results in Fig. 5, S3, S4

• switching_world_classifier.ipynb generates the evaluation dataset used to evaluate the classifier performance in Fig. 5f.

• src.run_multi_hmm_fits.py is run to generate the HMM fits for each animal.