Spatium documentation

Welcome to Spatium documentation. All the code can be found from https://github.com/ploughhh/Spatium.

Spatium is a foundation model for spatial proteomics that enables representation learning, cell type annotation, spatial niche analysis, protein imputation, multimodal inference, and downstream fine-tuning across diverse spatial proteomics datasets.

For information about the model architecture, pretraining datasets, benchmarking results, and citation information, please refer to the Introduction page.

To install Spatium and configure the required environment, please refer to the Installation page.

For examples of protein tokenization and data preprocessing, please refer to the Tokenizer page.

We also provide complete analysis workflows, covering the process from input spatial proteomics data to downstream model outputs. These tutorials will guide users through practical applications of Spatium. Please refer to the Spatium Tutorials page.

Contents:

• Introduction
• Installation
• Tokenizer
• Spatium Fine-tuning Training Example
• Spatium Tutorials