jtfouquier / explana

Workflow to generate interactive html feature selection report for longitudinal and cross-sectional microbiome studies

Copyright 2023-2024 Regents of the University of Colorado. All rights reserved.

Authored by Jennifer Therese Fouquier

Free for academic use. For other licensing information and user guide can be found at explana.io

Please contact me with any issues, suggestions or questions at [email protected]

EXPLANA: A user-friendly workflow for EXPLoratory ANAlysis and feature selection in longitudinal and cross-sectional microbiome studies

EXPLANA was developed to streamline identification of important features that relate to response variables in longitudinal, and cross-sectional, microbiome studies.

Response variables can be numerical or binary categorical and input features can be numerical or categorical. If order of categorical value changes impacts a response, this will be identified.

Features are ranked by their importance, and feature impact on response is additinoally provided. An interactive .html report is generated for each analysis that visually and textually summarizes results.

This is an exploratory method and this should be made clear when communicating results.

Install

Source code

Run the command:

git clone [email protected]:JTFouquier/explana.git

Next, navigate to directory with Snakefile and work from there.

---

Create Conda environment using Mamba

You must be inside the directory with Snakefile to create the explana environment and run your analyses.

You will need to install Mambaforge (download Mambaforge here)

mamba env create -f conda_envs/environment.yaml -n explana

Activate the explana conda environment. This is needed to ensure specific software versions are used.

conda activate explana

---

Test install

Run EXPLANA with small install demo. If possible, use more cores (I use 6).

snakemake --configfile config/config-install-test.yaml --cores 2 --use-conda

Open html report (or double click it from within directory)

open workflow-results/EXPLANA-install-test/EXPLANA-report.html

To retest, remove output folder.

To remove and retest (rm -r removes a folder and cannot be undone):

rm -r workflow-results/EXPLANA-install-test/

If workflow partially runs, rerun snakemake command and Snakemake will resume analysis.

---

FAQs and user manual coming soon

Useful Reading

• https://towardsdatascience.com/boruta-shap-an-amazing-tool-for-feature-selection-every-data-scientist-should-know-33a5f01285c0
• https://github.com/Ekeany/Boruta-Shap
• https://shap.readthedocs.io/en/latest/
• https://christophm.github.io/interpretable-ml-book/shap.html