Input data and analyzed data of "Topology of synaptic connectivity constrains neuronal stimulus representation (...)"

This dataset contains the input data, as well as the analyzed data that our preprint

Topology of synaptic connectivity constrains neuronal stimulus representation, predicting two complementary coding strategies

to be found on bioRxiv is based on. The input data (input_data.zip) contains everything that is needed to run the full analysis pipeline from start to the generation of the figures found in the manuscript. However, some of the analysis steps can be computationally heavy, so we also provide the output of these expensive steps, that can be simply used in conjunction with jupyter notebooks (notebooks.zip) to generate the figures.

Overview

An overview image can be found here

Blue squares denote input / output files (that are part of this dataset). Grey circles denote steps of the analysis pipeline (that are implemented in the github repository). Red rectangles denote configuration files (that are part of this dataset and also in the github repository).

Contained file types and their structure

Here, we provide four types of files. Configuration files specify analysis parameters and define the expected locations of the data files. Input files are the inputs into the analysis pipeline. Analyzed files are the outputs of said pipeline. Finally, we provide a number of jupyter notebooks that use the analyzed files to generate the manuscript figures. If you want to re-run the entire analysis pipeline, you need the code and configuration files from the repository, the input files and notebooks; the analyzed files will be generated as you run the pipeline. For information how to run this, refer to the readme. If you only want to generate the figures, you still need the code and configuration files from the repository, as it contains a package related to reading the result files; further, you need the analyzed files in addition to the input files. Of course, you can also run parts of the analysis pipeline and download the outputs for the rest.

To run everything smoothly, the files have to be placed into the expected file structure. You can look up and configure the file structure in the configuration files. Below, we describe the default layout, which is very simple (root is where you placed the code from our repository and can be any location on your file system):

• Configuration files: Part of the repository. Placed into root/working_dir/configs
• Input data: Place into root/working_dir/data, then unzip in place
  • input_data.zip -- Input data. Contains details on the model used in the manuscript and the output (spike times) of the simulation described in the manuscript. Within the file:
    • For details, see readme
• Analyzed data: Place into root/working_dir/data, then unzip in place
  • classifier_features_results.zip -- Output of the "classifier" step. Results of stimulus classification on the data in features.zip
  • classifier_manifold_results.zip -- Output of the "classifier" step. Results of stimulus classification on the data in extracted_components.zip
  • community_database.zip -- Output of "gen_topo_db". Various topological parameters related to the close neighborhood of neurons in the model
  • extracted_components.zip -- Output of "manifold_analysis". Results of factor analysis on the spike times in the input_data
  • features.zip  -- Output of "topological_featurization". A new dimensionality reduction method we introduce in the manuscript
  • split_spike_trains.zip  -- Output of "split_time_windows". The spike trains, split into time windows that are the responses to individual stimuli injected in the simulation
  • structural_parameters.zip -- Output of "Structural tribe analysis". Values for the topological parameters in community_database.zip associated with the neuron samples specified in tribes.zip
  • structural_parameters_vol.zip -- Output of "Structural tribe analysis". Same as above, but for volumetric neuron samples.
  • triads.zip -- Output of "Triad-counts". Over- and under-expression of triad motifs in the samples in tribes.zip.
  • tribes.zip -- Output of the step "sample_tribes". Volumetric and neighborhood-type neuron samples that are then analyzed further.
• Notebooks: Place into root/notebooks and unzip in place
  • notebooks.zip -- Jupyter notebooks. Run them to generate the figures in the manuscript.

Updates:

v1.1.0 (2020/12/11): Added some additional control cases to the results for figure 7. These results will probably not be updated on bioRxiv, but go into the submission to a journal.

Note (2021/09/20): This dataset contains the file "tribes.zip" that holds specifications of various neuron samples. "Tribe" is the term we used for samples of a neuron and its entire graph-theoretical neighborhood in an early version of the manuscript. We have since been alerted that this is unfortunate and non-inclusive terminology and have updated the manuscript to use the term "neighborhood" instead. Yet, we have decided not to reflect this change in the source code and data files at this point, as the likelihood to break something is too high. We apologize for any confusion or offense given.