cochlea-positional-information

To reproduce the results, do the following:

1. Install the Julia programming language (tested with Julia 1.7.1). MATLAB is also needed for mutual information calculations as implemented here (tested with MATLAB 2020a). This was developed on Windows 10.
2. Retrieve the code and data by downloading and extracting this archive file (~35 GB).
3. To activate the project environment and retreive the environment's dependencies from the package registry, open a Julia REPL (terminal), navigate to the repo directory, and do:

julia> using Pkg
julia> Pkg.activate(".")
julia> Pkg.instantiate()

4. To reproduce figures, navigate to the scripts/ directory and open each file as desired. This is easiest using Visual Studio Code using the Julia extension. Press ALT+ENTER to run the full script, which will load the relevant dataset(s). Ensure that the Julia session is running from the scripts/ directory. The option to save the figure as an image file may be un-commented at the bottom of each script. Note that some figures may be more efficiently viewed by saving rather than rendering in the plots pane.

Full reproduction should take less than 2 hours depending on network conditions and hardware specifications.

The codes required for preprocessing and analysis are in src/:

• analysis.jl holds a function for generating decoding maps for single and jointly measured expression profiles.
• bmp-model_fst.jl contains the mechanistic BMP model with Follistatin.
• bmp-model_no-fst.jl contains the BMP model without Follistatin.
• build-cochlea-dataframes.jl specifies the ETL settings for
• data_preprocessing.jl holds several functions related to preprocessing of raw image quantification files.
• extrapolatemiestimates.m debiases mutual information estimates from a limited dataset through bootstrapping and extrapolation.
• histcountsn.m creates N-dimensional histograms for mutual information calculations.
• midirectestimaten.m calculates naive estimates for mutual information to feed to the extrapolation function.
• mutual_information_script.m calls the above functions to estimate the mutual information in loaded profile datasets.
• synth_profile_mi_script.m estimates mutual information related to Figure 5.
• utils.jl contains various helper functions for preprocessing and ETL.

The data is organized into subdirectories under data/:

• exp_pro/ contains processed experimental data including consolidated dataframes and processed images.
• exp_raw/ contains raw image data.
• sims/ contains BMP model simulation outputs and synthetic profiles related to Figure 5.

This work is licensed under a Creative Commons Attribution 4.0 International License for data. and a BSD 3-Clause License for code.