MetaboNexus User Guide

From installation to publication-ready figures, this guide covers the entire workflow.

Part 1: The Preprocessing Tool 🧼

This essential first step cleans and prepares raw spatial metabolomics data, primarily targeting the `.h5ad` format. It addresses the critical challenge of distinguishing true tissue signals from the background matrix or slide surface.

1

Tissue Detection

MetaboNexus employs a ResNet34-based deep learning model, trained on a vast library of Total Ion Current (TIC) images, to automatically predict and mask the tissue area. Because no model is perfect, a crucial Manual Override feature is included. This provides an intuitive brush-and-eraser interface, allowing you to paint or remove spots directly on the TIC image for final control over the tissue mask.

2

Marker Metabolite & Drug Filtering

You can optionally refine your dataset to include only the most biologically relevant metabolites. This feature performs a T-test to identify and retain molecules that are significantly more abundant in tissue vs. background, based on your defined fold-change and p-value thresholds. Furthermore, you can automatically filter out known drug compounds using an integrated DrugBank database or provide a custom list of molecules (e.g., contaminants) to exclude.

3

Output

Once preprocessing is complete, you can save the cleaned data in your format of choice, including `.csv`, `.xlsx`, `.txt`, `.h5`, or the analysis-ready `.h5ad` format.

Part 2: The Main Analysis Workflow 🚀

With clean data, you can proceed to the main application. The interface is organized into four sequential panels that guide you through the process.

1

Define File Groups

This is the primary data input step where you group your data files by experimental condition, tissue type, or timepoint. The application accepts `.h5ad`, `.h5`, `.csv`, and `.xlsx` files.

2

Select Analyses

Choose from the comprehensive suite of built-in analysis modules. You can select multiple analyses to run in a single batch.

3

Configure Plot Options

For each analysis you've selected, choose the specific visualizations you want the application to generate.

4

Final Configuration & Execution

This panel gives you granular control over the entire run. You can adjust all analysis parameters, apply statistical pre-filters (e.g., ANOVA, T-test) to focus on significant metabolites, and configure complex multi-group comparisons using the Paired Comparison Mode.

Installation & System Requirements 💻

MetaboNexus is provided as a standalone executable, meaning no Python or library installation is required by the user.

Supported Operating Systems

  • Windows: Windows 10/11
  • macOS: Officially supported on Intel & Apple Silicon (M1/M2/M3) chips
  • Linux: Not officially supported at this time

First-Time Setup

Note: On the first run, the application will require an internet connection to automatically download necessary large assets. This includes the tissue prediction model and the enrichment analysis databases. This is a one-time process.

How to Cite ✍️

We are currently in the process of publishing a manuscript describing MetaboNexus. Until the peer-reviewed paper is available, if you use this application in your research, please cite the official website:

https://metabonexus.tech