spEnhance: A Computational Framework for Trustworthy Super-Resolution Enhancement in Spatial Omics

spEnhance is a computational framework for trustworthy super-resolution enhancement in spatial omics data. It starts from spot-level spatial transcriptomics (ST) data, paired H&E image and single cell reference, and reconstructs fine-grained, trustworthy super-resolution spatial gene expression.

Installation

Install the latest version from https://github.com/dsong-lab/spEnhance

Documentation

OVERVIEW

• About spEnhance
• spEnhance Workflow

TUTORIAL

• Image Preprocessing
  • Image rescaling
  • Image padding
• Histology Feature Extraction
  • Global histological feature extraction using HIPT
  • Fine-grained histological feature extraction using UNI
  • Integration of global and fine-grained histological features
• Validation Set Construction
• Single-cell Reference Guided Gene Expression Embedding
  • Cell-type deconvolution using RCTD
  • Pixel-level cell type prediction
  • Gene expression feature assignment
• Spatial Co-expression Modules Identification
• Trustworthy prediction of super-resolution spatial gene expression
  • Impute super-resolution spatial gene expression using GCN
  • Error quantification to prevent model overfitting