AgingDrugOmics

AgingDrugOmics uses both metabolomic and proteomic plasma data to predict whether a 12-month-old UM-HET3 male mouse has been given an anti-aging intervention. The machine-learning model employed is XGBoost, and the analysis requires both R and Python.

Note: Access to the raw data (Step 1) will be provided upon manuscript acceptance.

---

Overview

• Goal: Discriminate between control and slow-aging (anti-aging treated) mice using metabolomic and proteomic datasets.
• Highlights:
  • Normalizes and combines multiple omics datasets.
  • Performs 10k cross-fold validation.
  • Tests on novel interventions excluded from training data.
  • Generates SHAP values and visualizations for model interpretability.

This repository accompanies the manuscript titled:
“[Discrimination of Normal from Slow-Aging Mice by Plasma Metabolomic and Proteomic Features – Link will be added upon publication]”

---

System Requirements

• Operating System:

  • Tested on Windows 11 (version 24H2)

• Software:

  • R (Recommended version: 4.0+)
  • Python (Recommended version: 3.8+)
  • RStudio and/or Anaconda (Jupyter Lab) for notebook execution

---

Dependencies

A detailed list of required packages for each step of the pipeline is available in the Package_Info folder.

---

How to Run the Code

1. Download the Repository

   • Click the green “Code” button and select “Download ZIP” or clone via Git:

     git clone https://github.com/BrettonB/AgingDrugOmics.git

   • Navigate into the directory:

     cd AgingDrugOmics

2. Prepare Your Environment

   • Python: install anaconda (https://www.anaconda.com/products/navigator) and start Jupyter Lab and install the dependencies in Package_Info folder.
   • R: Install R studio (https://posit.co/download/rstudio-desktop/) and install the dependencies in Package_Info folder.

3. Follow the Steps Sequentially
   Each step in the analysis pipeline is labeled Step#... (from Step 1 through Step 13).

   • Start with Step 1 and proceed in numerical order.
   • Within each step’s folder, there is either an .Rmd file (R Markdown) or an .ipynb file (Jupyter Notebook).
   • Open and run the corresponding .Rmd or .ipynb in RStudio or Jupyter Lab. These files will orchestrate the associated script files.

4. Check the Output

   • All relevant outputs (figures, intermediate data, final results) will be saved in the Paper_Figures folder (and subfolders within it).
   • Log or console messages will guide you on any required input data formats or additional steps.

---

High-Level Code Functionality

1. Input Data

   • Metabolomic data for control vs. anti-aging–treated mice (named metabolites, not raw peaks).
   • Proteomic data for control vs. anti-aging–treated mice (named proteins, not raw peaks).
   • The abovde datasets include metadata describing intervention groups, sex, sample IDs, and the metabolites.

2. Data Processing & Modeling

   • Normalization and Integration: Merges metabolomic and proteomic data into a unified dataset.
   • Cross-Fold Validation: Performs 10K cross-fold splits for robust performance estimation.
   • Novel Intervention Testing: Excludes one intervention group from training to assess predictive performance on truly novel data.
   • Feature Importance (SHAP): Calculates SHAP values for model interpretability.

3. Output

   • Normalized Datasets: Cleaned and combined metabolomic/proteomic matrices.
   • Prediction Results: statistics for the cross-fold validation and novel-intervention tests.
   • SHAP Interpretations: Feature importance rankings and visual explanations for model decisions.
   • Visualizations: Plots and figures demonstrating performance, data distributions, and SHAP results.

---

Future Updates

• Raw Data: Will be uploaded following the acceptance of the associated manuscript.
• Paper Link: The README will be updated once the paper is published and publicly accessible.

---

Contact

For questions or collaboration inquiries, please reach out via the GitHub Issues section or email [brettonb@umich.edu].

---