Analysis

The final component of the HSNPipeline is a template structure for analyzing the data.

Overview

Once the data is pre-processed and organized into a standardized format, the next task is to analyze it! While the analysis of any given project will be specific to the task and research questions, this section of the pipeline provides a structured layout for organizing these analyses.

Resources

As with other sections of this pipeline, the template itself does not implement any key tooling, but rather provides an organized structure for human single-neuron analyses, that uses existing analysis tools. The key resources and tools are briefly described in this section.

As well as the below mentioned tools, data analysis in the HSNPipeline requires the hsntools module.

SpikeTools

spiketools is a Python based module for analyzing spiking data, including functionality that is specifically designed for analyzing human data.

For more information on spiketools, see:

• The spiketools documentation
• The spiketools repository
• The spiketools paper

AnalyzeTEMPLATE

The AnalyzeTEMPLATE provides a template structure for organizing the analysis of a project examining human single-neuron data.

The suggested structure of the template includes:

• a sub-module for custom analysis code
• a set of notebooks for developing, exploring, and demonstrating analyses
• an organized set of script templates for running analyses across the dataset

Typically, analyses can be organized at different levels, for example:

• group level analyses, that summarize / analyze data across the whole group
• session level analyses, that run analyses across each session of data (e.g. behavioural results)
• unit level analyses, that run a set of analyses on each neuron

In doing these analysis, the recommended approach is to organize analyses, each with their own visualizations, and then create reports that collect together all the visualizations for each analysis.

Run Procedures

The suggested process for developing and applying analyses is as follows:

• analyses can be developed in notebooks dedicated to each analysis
• as analyses mature, any custom code can be organized and moved into the local code folder
• once analyses are ready to run across the data, they can be added / converted into scripts
• at this point, the entire set of analyses can be executed be (re-)running the scripts
• additional notebooks can be created to collect across analyses, and create final figures

For example, the following would re-run all analyses:

# Run analyses across the group / sessions / all neurons
python scripts/group_analysis.py
python scripts/session_analysis.py
python scripts/unit_analysis.py