A repository containing the code accompanying the research paper "Neuronal travelling waves explain rotational dynamics in experimental datasets and modelling" by Kuzmina E., Kriukov D., Lebedev M., 2024, published in scientific reports. Preprint: bioRxiv
This repository contains code from a research paper focused on the analysis of neural activity data. Our work delves into the exploration and characterization of rotational dynamic prevalent in various neural datasets. We introduce a mathematical framework, informed by our research, designed to assess and quantify the "rotationess" of datasets. This framework leverages Gyration Numbers, a complex-valued metric derived from the eigenvalue decomposition of the differential covariance matrix of the data. The resulting Gyration Plane facilitates the comparison and simultaneous analysis of multiple datasets.
The brain areas that were explicitly studied with rotational dynamics approach in rhesus monkeys, humans and rodents.
- Clone the repository:
git clone https://github.com/NevVerVer/neural-dynamics-gyration.git- Navigate to the repository directory:
cd neural-dynamics-gyration- Install the required packages:
pip install -r requirements.txt-
Run script that downloads and unzip data:
bash prepare_data.shorbash prepare_preproc_datasets.sh. Or download archive with datasets manually from here. You can also download datasets from the original source from here ... # TODO link to table.dataset.zipcontains raw data, as it is available at origin sourse.preprocessed2h5.zipcontains already prerocessed datasets, saved to .h5 file as dictionary. -
Add path to downloaded datasets to
datasets_config.pyfile for convinience.
Place your data in ./datasets/ directory or execture script prepare_data.sh to download data used in the study.
If you would like to work with datasets used in study, you can see how to pre-process and save datasets to h5 file in notebook datasets_analysis.ipynb. If using other dataset, you can use parent class NeuralDataset (from utils/datasets.py) and add specific methods to load data and pre-process.
-
datasets_analysis.ipynb- contains tutorial on how to use special class created for working with datasets. It is easy modifiable and allows to use all visualization functions that were used in the study. Have code that was used to render Fig. S2-S3-S4 and Fig. 4 from our paper. -
gyration_plane_tutorial.ipynb- contains tutorial on how to use and interpret Gyration Number concept. How to plot datasets in Gyration Plan and compare them. Also, contains bonus explanation of Curvature concept. -
traveling_wave_model.ipnyb- Have code that was used to render Fig. 3, 5 and Fig. S5-S6-S7 from our paper.
In short - Gyration Plane is an allocation of neural population dynamics datasets in the space spanned by real and imaginary parts of their first pair of complex-conjugated eigenvalues normalized by the spectrum power.
The normalized imaginary part measures the structural rotation component in the data and the normalized real part measures the inflation/deflation component of the data.
This visualization approach provides a holistic view of the rotational dynamic problem allowing a comparison of different datasets with each other in the model-agnostic (no fitting as in jPCA) manner not requiring a dimensionality reduction procedure.
How different rotational activity locates in Gyration Plane
In utils/shuffling.py you can find implementation of Shuffling Procedures that are widely used for validation of "importance" of rotational dynamics in data.
We implemented three types of shufflings from original Churchland et al., and CMPT shuffling from Micaels et al. Feel free to use them!
If you use the code or the findings from our paper, please cite:
Kuzmina, E., Kriukov, D., & Lebedev, M. (2024). Neuronal travelling waves explain rotational dynamics in experimental datasets and modelling. Scientific Reports, 14(1), 3566.(doi: https://doi.org/10.1038/s41598-024-53907-2)
@article{kuzmina2024neuronal,
title={Neuronal travelling waves explain rotational dynamics in experimental datasets and modelling},
author={Kuzmina, Ekaterina and Kriukov, Dmitrii and Lebedev, Mikhail},
journal={Scientific Reports},
volume={14},
number={1},
pages={3566},
year={2024},
publisher={Nature Publishing Group UK London}
}If you use the code or the findings from our preprint, please cite:
Kuzmina, E., Kriukov, D., & Lebedev, M. (2023). On the Rotational Structure in Neural Data. bioRxiv, 2023-09. (doi: https://doi.org/10.1101/2023.09.11.557230)
@article{kuzmina2023rotational,
title={On the Rotational Structure in Neural Data},
author={Kuzmina, Ekaterina and Kriukov, Dmitrii and Lebedev, Mikhail},
journal={bioRxiv},
pages={2023--09},
year={2023},
publisher={Cold Spring Harbor Laboratory}
}For any questions or clarifications, please reach out to: ekaterina.kuzmina@skoltech.ru
| Name & Authors | Year | Brain Areas | Behavior | Source | Original Paper |
|---|---|---|---|---|---|
| "Neural Population Dynamics During Reaching", Churchland et al. | 2012 | primary motor cortex (M1), dorsal premotor cortex (PMd) | straight hand reaches, curved hand reaches | ... | link |
| "Local field potentials reflect cortical population dynamics in a region-specific and frequency-dependent manner", Gallego et al. | 2022 | primary motor cortex (M1), dorsal premotor cortex (PMd), primary somatosensory cortex (area 2) | delay centre-out reaching task using a manipulandum | dryad | link |
| "Neural Population Dynamics in Motor Cortex are Different for Reach and Grasp", Suresh et al. | 2020 | primary motor cortex (M1), somatosensory cortex (SCx) | isolated grasping task, center-out reaching task | dryad | link |
| "Rotational dynamics in motor cortex are consistent with a feedback controller", Kalidindi et al. | 2021 | premotor and primary motor cortex (MC), primary somatosensory cortex (areas 1, 2, 3a, 5) | delayed center-out reaching task, posture perturbation task | dryad github | link |
| "Context-dependent Computation by Recurrent Dynamics in Prefrontal Cortex", Mante et al. | 2013 | prefrontal cortex (PFC): arcuate sulcus in and around the frontal eye field (FEF) | context-dependent 2-alternative forced-choice visual discrimination task (perceptual decision-making) | iniuzhch | link |
Special thanks to @nosmokingsurfer for their valuable feedback and suggestions.
Thanks to @bantin for the jPCA implementation.
We welcome contributions to this repository. If you're found errors in code or experiments, please open an issue to discuss your ideas.
This project is licensed under the MIT License - see the LICENSE file for details.