Repository for project associated with manuscript "Neural circuit models capture human subjective auditory perception," also known as prophecy.
All code and analyses for prophecy are writen in Python and require Python >=3.6 to run.
It has the following required dependences:
- numpy
- scipy
- pandas
- mne
- mne-bids
- neurodsp
The accompanying jupyter notebooks are used for visualization and statistical analysis and have the following additional dependences:
- matplotlib
- seaborn
- pingouin
- pyqt5
The Anaconda distribution is recommended to manage these requirements.
To install and run code associated with prophecy, clone this repository and install the requirements using the following commands:
git clone https://github.com/voytekresearch/prophecy.git
pip install -r requirements.txt
Installation with a conda environment is highly recommended. Typical install time should be <5 minutes.
To demonstrate the analysis, simulated sEEG data is provided in 00-data. To run the main analysis on simulated data, use the following commands in a python/conda environment with the required dependencies:
python prophecy_main.py
Run time:
To run the permutation test:
python prophecy_permut.py
Run time:
To run the parameter sweep (Supp. Fig. 6):
python parameter_sweep.py
Run time:
NOTE Results in this demo are generated with simulated data, and will not replicate manuscript results. To replicate results described in the manuscript, run any of the provided notebooks (.ipynb). Although raw sEEG data cannot be provided at this time, .csv filees with real analysis outputs are provided in 00-data.
To run a similar analysis on your own data, we recommend using the script prophecy_start_here.py. This script provides a simplified analysis, where a hypothesized spike probability signal (like the SAM y-unit described in the manuscript) is used to simulate a putative local field potential, which is then correlated to raw sEEG data on a single trial.