Analogous cycles

This repository contains code accompanying the following paper: Yoon et al 2024 "Tracking the topology of neural manifolds across populations".

• One should first download & install Julia. The code has been tested on Julia version 1.10.4.
• We recommend running the code on a compute server rather than a laptop.
• Due to the large size of data, not all data files are included in this repository. Please contact Iris Yoon (hyoon@wesleyan.edu) for copies of the data.
• The analogous cycles code is written in Julia. However, some of the preprocessing steps are done in Python. Each notebook clarifies which language one should use.
• We use the phrase analogous cycles and analogous bars interchangably.

Quick-start: implementation of analogous cycles

• Download & install Julia.
• The "examples" directory contain example notebooks illustrating how to compute analogous cycles.
• When running the analogous cycles code, it is important that one activates the virutal environment associated with this project. See the following sections for instructions on using Jupyter notebook and activating the proper virtual environment.

Instructions: Running Jupyter notebook from Julia REPL

• Open Julia REPL
• Using the cd command, navigate to the root of this directory.
• Activate & instantiate the virtual environment as follows.

using Pkg
Pkg.activate( "/env/.")
Pkg.instantiate()

• Open a notebook using the following command.

Pkg.build("IJulia")
using IJulia
notebook()

• You can now run the Jupyter notebooks in this repository.

Instructions on running code on your data

When running the analogous cycles script on your own data, please make sure to include the following command.

using Pkg
Pkg.activate( "/env/.")
Pkg.instantiate()

include("src/Eirene_var.jl")
include("src/analogous_bars.jl")

using .Eirene_var
using .analogous_bars

Running the code on the provided datasets

• See theanalysis directory.
• We recommend running the code in the experimental_visual or simulation_navigation directories, as the computation can take a long time for the dataset in simulation_visual.
• For simulation visual, we recommend running the code on a server.