somata

Github: https://github.com/mh105/somata

State-space Oscillator Modeling And Time-series Analysis (SOMATA) is a Python library for state-space neural signal processing algorithms developed in the Purdon Lab. Basic state-space models are introduced as class objects for flexible manipulations. Classical exact and approximate inference algorithms are implemented and interfaced as class methods. Advanced neural oscillator modeling techniques are brought together to work synergistically.

---

Table of Contents

• Requirements
• Install
• Basic state-space models
  • StateSpaceModel
  • OscillatorModel
  • AutoRegModel
  • GeneralSSModel
• Advanced neural oscillator methods
• Authors
• Citation
• License

---

Requirements

somata is built on numpy arrays for computations. joblib is used for multithreading. Additional dependencies include scipy, matplotlib, cmdstanpy, and spectrum. The source localization module also requires pytorch and MNE-python.

• Full package requirements for each release will be updated in the requirements-*.txt files. The pyproject.toml file is specified to dynamically retrieve the metadata of dependencies for setuptools during pip install to verify that runtime dependent packages of compatible versions have been installed. When pip install is used, missing dependencies will be fetched from Python Package Index (PyPI) and installed.

• For development or installing somata into a new conda environment, requirements-*.txt files can also be passed to conda create via the --file directive to create a new conda environment with all and only the required packages installed in the new conda environment. When conda create or conda install is used, missing dependencies will be fetched from conda channels and installed.

Install

$ pip install somata

or

$ conda install somata -c pytorch -c conda-forge

conda-forge channel (recommended)

While pip install usually works, an alternative way to install somata is through the conda-forge channel, which utilizes continuous integration (CI) across OS platforms. This means that conda-forge packages are more compatible with each other compared to PyPI packages installed via pip by default. If pip install somata fails to resolve some dependencies, the conda-forge somata feedstock can be used to install somata. If you have not installed CmdStan on your OS before, it is easier to install from the conda-forge channel if you need to use the Phase Amplitude Coupling Estimation module. On Windows, it is also easier to use conda install due to a need to build spectrum during pip install that requires Microsoft Visual C++ 14.0 or greater.

torch requirement

If the torch dependency is not resolved correctly for your OS (such as installed the cpu-only version when GPU processing is needed), first install pytorch manually in a conda environment that you want to install somata in, and then rerun either of the above two lines to install somata. Please be aware of a common mixup that PyTorch is distributed as torch on PyPI but as pytorch on conda-forge. If using conda to install, Windows OS needs to download pytorch from the pytorch channel, as win-64 is not built on the conda-forge channel.

(For development only)

• Fork this repo to personal git

  How to: GitHub fork

• Clone forked copy to local computer

  How to: GitHub clone

• Install conda

  Recommended conda distribution: Miniforge3

  Apple silicon Mac: choose Miniforge3 native to the ARM64 architecture instead of Intel x86.

• Create a new conda environment

  You may also directly install somata in an existing conda environment by skipping this step.

  $ cd <repo root directory with pyproject.toml>
$ mamba create -n somata -c pytorch -c conda-forge --file requirements/requirements-core.txt --file requirements/requirements-dev.txt
$ mamba activate somata

• Install somata as a package in development mode

  $ cd <repo root directory with pyproject.toml>
$ pip install -e . --config-settings editable_mode=compat

  What is: Editable Installs

• Configure IDEs to use the conda environment

  How to: Configure an existing conda environment

Some notes on package dependency requirements

When somata is installed into an existing conda environment, unmet dependencies are automatically searched, downloaded, and installed from the same repository of packages requested to provide the somata build distribution and its dependencies (from PyPI with pip install somata or from pytorch and conda-forge channels with conda install somata -c pytorch -c conda-forge).

The need for specifying dependencies arises in multiple ways for Python. There are non-negligible complexities due to the existence of different build backends as well as different Python dependency management and packaging tools such as pdm, poetry, pip, conda, etc. One modern standard is to use a declarative config pyproject.toml file for package